diff --git a/mimic3_make_harvard_sentences.py b/mimic3_make_harvard_sentences.py index 19956c7ee5aa9c005170eaee69490da57b14469a..845f88fe7e415b9c553634da9ccc1e55ccee96bb 100644 --- a/mimic3_make_harvard_sentences.py +++ b/mimic3_make_harvard_sentences.py @@ -77,6 +77,21 @@ list_voices = [ # ================================================== INTERFACE MODELS + + + + + + + + + + + + + + + LABELS = [ 'arousal', 'dominance', 'valence', # 'speech_synthesizer', 'synthetic_singing', @@ -131,10 +146,77 @@ teacher_cat.forward = types.MethodType(_infer, teacher_cat) -# Audioset & ADV +# ===================[:]===================== Dawn +def _prenorm(x, attention_mask=None): + '''mean/var''' + if attention_mask is not None: + N = attention_mask.sum(1, keepdim=True) # here attn msk is unprocessed just the original input + x -= x.sum(1, keepdim=True) / N + var = (x * x).sum(1, keepdim=True) / N + + else: + x -= x.mean(1, keepdim=True) # mean is an onnx operator reducemean saves some ops compared to casting integer N to float and the div + var = (x * x).mean(1, keepdim=True) + return x / torch.sqrt(var + 1e-7) + +from torch import nn +from transformers.models.wav2vec2.modeling_wav2vec2 import Wav2Vec2PreTrainedModel, Wav2Vec2Model +class RegressionHead(nn.Module): + r"""Classification head.""" + + def __init__(self, config): + + super().__init__() + + self.dense = nn.Linear(config.hidden_size, config.hidden_size) + self.dropout = nn.Dropout(config.final_dropout) + self.out_proj = nn.Linear(config.hidden_size, config.num_labels) + + def forward(self, features, **kwargs): + + x = features + x = self.dropout(x) + x = self.dense(x) + x = torch.tanh(x) + x = self.dropout(x) + x = self.out_proj(x) + + return x + + +class Dawn(Wav2Vec2PreTrainedModel): + r"""Speech emotion classifier.""" + + def __init__(self, config): + + super().__init__(config) + + self.config = config + self.wav2vec2 = Wav2Vec2Model(config) + self.classifier = RegressionHead(config) + self.init_weights() + + def forward( + self, + input_values, + attention_mask=None, + ): + x = _prenorm(input_values, attention_mask=attention_mask) + outputs = self.wav2vec2(x, attention_mask=attention_mask) + hidden_states = outputs[0] + hidden_states = torch.mean(hidden_states, dim=1) + logits = self.classifier(hidden_states) + return logits + # return {'hidden_states': hidden_states, + # 'logits': logits} +dawn = Dawn.from_pretrained('audeering/wav2vec2-large-robust-12-ft-emotion-msp-dim').to(config.dev).eval() +# ======================================= + + + + + -# audioset_model = audonnx.load(audmodel.load('17c240ec-1.0.0'), device='cuda:0') -adv_model = audonnx.load(audmodel.load('90398682-2.0.0'), device='cuda:0') def process_function(x, sampling_rate, idx): '''run audioset ct, adv @@ -154,7 +236,7 @@ def process_function(x, sampling_rate, idx): # logits_audioset = audioset_model(x, 16000)['logits_sounds'] # logits_audioset = logits_audioset[:, [7, 35]] # speech synthesizer synthetic singing # -- - logits_adv = adv_model(x, 16000)['logits'] + logits_adv = dawn(torch.from_numpy(x).to(config.dev)).cpu().detach().numpy() #['logits'] cat = np.concatenate([logits_adv, # _sigmoid(logits_audioset), @@ -169,7 +251,7 @@ interface = audinterface.Feature( # process_func_args={'outputs': 'logits_scene'}, process_func_applies_sliding_window=False, win_dur=7.0, - hop_dur=4.0, + hop_dur=40.0, sampling_rate=16000, resample=True, verbose=True, @@ -297,7 +379,7 @@ for _id, _voice in enumerate(list_voices): total_audio_mimic3 = [] total_audio_styletts2 = [] ix = 0 - for list_of_10 in harvard_individual_sentences[:1000]: # 77 + for list_of_10 in harvard_individual_sentences[:4]: # 77 text = ' '.join(list_of_10['sentences']) diff --git a/style_vector/en_UK_apope.wav b/style_vector/en_UK_apope.wav new file mode 100644 index 0000000000000000000000000000000000000000..b4f7c11403e011a6eba7301e8e45af6df0cf3a8d Binary files /dev/null and b/style_vector/en_UK_apope.wav differ diff --git a/style_vector/en_US_cmu_arctic_aew.wav b/style_vector/en_US_cmu_arctic_aew.wav new file mode 100644 index 0000000000000000000000000000000000000000..9a61735b23953c4178e48504d7970f646163c459 Binary files /dev/null and b/style_vector/en_US_cmu_arctic_aew.wav differ diff --git a/style_vector/en_US_cmu_arctic_ahw.wav b/style_vector/en_US_cmu_arctic_ahw.wav new file mode 100644 index 0000000000000000000000000000000000000000..f445dda4bd9cfd43d641a344d1844fb847506514 Binary files /dev/null and b/style_vector/en_US_cmu_arctic_ahw.wav differ diff --git a/style_vector/en_US_cmu_arctic_aup.wav b/style_vector/en_US_cmu_arctic_aup.wav new file mode 100644 index 0000000000000000000000000000000000000000..4aedbe4e1bf031a314a9eeaad4033d1ce5d0d067 Binary files /dev/null and b/style_vector/en_US_cmu_arctic_aup.wav differ diff --git a/style_vector/en_US_cmu_arctic_awbrms.wav b/style_vector/en_US_cmu_arctic_awbrms.wav new file mode 100644 index 0000000000000000000000000000000000000000..dbe41efca10db9f5dea51142b3d2a63c39a08cfb Binary files /dev/null and b/style_vector/en_US_cmu_arctic_awbrms.wav differ diff --git a/style_vector/en_US_cmu_arctic_axb.wav b/style_vector/en_US_cmu_arctic_axb.wav new file mode 100644 index 0000000000000000000000000000000000000000..3112edd92056ea3259e3d02d1a9abf888dbd2f11 Binary files /dev/null and b/style_vector/en_US_cmu_arctic_axb.wav differ diff --git a/style_vector/en_US_cmu_arctic_bdl.wav b/style_vector/en_US_cmu_arctic_bdl.wav new file mode 100644 index 0000000000000000000000000000000000000000..806eb32572734cce078484e475abac41a985cc19 Binary files /dev/null and b/style_vector/en_US_cmu_arctic_bdl.wav differ diff --git a/style_vector/en_US_cmu_arctic_clb.wav b/style_vector/en_US_cmu_arctic_clb.wav new file mode 100644 index 0000000000000000000000000000000000000000..bfa4f584252ac40ab917f14224b0c7f670783397 Binary files /dev/null and b/style_vector/en_US_cmu_arctic_clb.wav differ diff --git a/style_vector/en_US_cmu_arctic_eey.wav b/style_vector/en_US_cmu_arctic_eey.wav new file mode 100644 index 0000000000000000000000000000000000000000..21b39a1da3e5ed8c438879fb3ac627721c80b4ff Binary files /dev/null and b/style_vector/en_US_cmu_arctic_eey.wav differ diff --git a/style_vector/en_US_cmu_arctic_fem.wav b/style_vector/en_US_cmu_arctic_fem.wav new file mode 100644 index 0000000000000000000000000000000000000000..2cf13a935fd92a67aaaaffddb07b855d13163df9 Binary files /dev/null and b/style_vector/en_US_cmu_arctic_fem.wav differ diff --git a/style_vector/en_US_cmu_arctic_gka.wav b/style_vector/en_US_cmu_arctic_gka.wav new file mode 100644 index 0000000000000000000000000000000000000000..de973c4a5cfc4f222437018ef07c5d36d8c43ebd Binary files /dev/null and b/style_vector/en_US_cmu_arctic_gka.wav differ diff --git a/style_vector/en_US_cmu_arctic_jmk.wav b/style_vector/en_US_cmu_arctic_jmk.wav new file mode 100644 index 0000000000000000000000000000000000000000..f5766bf2bba082ebbb03c65afe7fe9406aea4942 Binary files /dev/null and b/style_vector/en_US_cmu_arctic_jmk.wav differ diff --git a/style_vector/en_US_cmu_arctic_ksp.wav b/style_vector/en_US_cmu_arctic_ksp.wav new file mode 100644 index 0000000000000000000000000000000000000000..8e0d415438cd2e495df5e0e8aa9648784ec2a0e5 Binary files /dev/null and b/style_vector/en_US_cmu_arctic_ksp.wav differ diff --git a/style_vector/en_US_cmu_arctic_ljm.wav b/style_vector/en_US_cmu_arctic_ljm.wav new file mode 100644 index 0000000000000000000000000000000000000000..71a993f47fd422c2fd394650134929388fb3319e Binary files /dev/null and b/style_vector/en_US_cmu_arctic_ljm.wav differ diff --git a/style_vector/en_US_cmu_arctic_lnh.wav b/style_vector/en_US_cmu_arctic_lnh.wav new file mode 100644 index 0000000000000000000000000000000000000000..1b00a1d3f5674378c1181d87fdf16446cff5fd83 Binary files /dev/null and b/style_vector/en_US_cmu_arctic_lnh.wav differ diff --git a/style_vector/en_US_cmu_arctic_rxr.wav b/style_vector/en_US_cmu_arctic_rxr.wav new file mode 100644 index 0000000000000000000000000000000000000000..904575732f818e973d8079649f168713bc46791a Binary files /dev/null and b/style_vector/en_US_cmu_arctic_rxr.wav differ diff --git a/style_vector/en_US_cmu_arctic_slp.wav b/style_vector/en_US_cmu_arctic_slp.wav new file mode 100644 index 0000000000000000000000000000000000000000..4edd88e69ee05f59bc7f0af69e1aace83d69c57f Binary files /dev/null and b/style_vector/en_US_cmu_arctic_slp.wav differ diff --git a/style_vector/en_US_cmu_arctic_slt.wav b/style_vector/en_US_cmu_arctic_slt.wav new file mode 100644 index 0000000000000000000000000000000000000000..26fac2d75244d8a79e3dc92fa255170c0af2fac3 Binary files /dev/null and b/style_vector/en_US_cmu_arctic_slt.wav differ diff --git a/style_vector/en_US_hifi-tts_6097.wav b/style_vector/en_US_hifi-tts_6097.wav new file mode 100644 index 0000000000000000000000000000000000000000..e44f41bd7c51ed74a5fd4bf546bd3b3fe2d6962c Binary files /dev/null and b/style_vector/en_US_hifi-tts_6097.wav differ diff --git a/style_vector/en_US_hifi-tts_9017.wav b/style_vector/en_US_hifi-tts_9017.wav new file mode 100644 index 0000000000000000000000000000000000000000..22efc568d97667a2f28ab7bba9d95c7712ccf975 Binary files /dev/null and b/style_vector/en_US_hifi-tts_9017.wav differ diff --git a/style_vector/en_US_hifi-tts_92.wav b/style_vector/en_US_hifi-tts_92.wav new file mode 100644 index 0000000000000000000000000000000000000000..c904a760b2633aed2f7b1e7122c5356c4cc472b8 Binary files /dev/null and b/style_vector/en_US_hifi-tts_92.wav differ diff --git a/style_vector/en_US_ljspeech.wav b/style_vector/en_US_ljspeech.wav new file mode 100644 index 0000000000000000000000000000000000000000..f85d1a155c2570f02df1128d95aac589491c193f Binary files /dev/null and b/style_vector/en_US_ljspeech.wav differ diff --git a/style_vector/en_US_m-ailabs_elliot_miller.wav b/style_vector/en_US_m-ailabs_elliot_miller.wav new file mode 100644 index 0000000000000000000000000000000000000000..9c62fc3c7c73e0d932604ab06d8dfaf3a5654b5d Binary files /dev/null and b/style_vector/en_US_m-ailabs_elliot_miller.wav differ diff --git a/style_vector/en_US_m-ailabs_judy_bieber.wav b/style_vector/en_US_m-ailabs_judy_bieber.wav new file mode 100644 index 0000000000000000000000000000000000000000..088f7b2bdeef6e6e1d9cbba4cb095895f4139e27 Binary files /dev/null and b/style_vector/en_US_m-ailabs_judy_bieber.wav differ diff --git a/style_vector/en_US_m-ailabs_mary_ann.wav b/style_vector/en_US_m-ailabs_mary_ann.wav new file mode 100644 index 0000000000000000000000000000000000000000..ed16ed72d33bd389ac97c712bb91d3c69ac71e74 Binary files /dev/null and b/style_vector/en_US_m-ailabs_mary_ann.wav differ diff --git a/style_vector/en_US_vctk_p225.wav b/style_vector/en_US_vctk_p225.wav new file mode 100644 index 0000000000000000000000000000000000000000..a060aa745c30b8dffd7c98004c6961922e4635af Binary files /dev/null and b/style_vector/en_US_vctk_p225.wav differ diff --git a/style_vector/en_US_vctk_p226.wav b/style_vector/en_US_vctk_p226.wav new file mode 100644 index 0000000000000000000000000000000000000000..ae7b99e8fdff6455294b94e4cab8e7b6824b8123 Binary files /dev/null and b/style_vector/en_US_vctk_p226.wav differ diff --git a/style_vector/en_US_vctk_p227.wav b/style_vector/en_US_vctk_p227.wav new file mode 100644 index 0000000000000000000000000000000000000000..f7c8910a79b44439a98439a4086d2d69df601676 Binary files /dev/null and b/style_vector/en_US_vctk_p227.wav differ diff --git a/style_vector/en_US_vctk_p228.wav b/style_vector/en_US_vctk_p228.wav new file mode 100644 index 0000000000000000000000000000000000000000..11bd485b313930b3b7d7f7543855b263123fc2e6 Binary files /dev/null and b/style_vector/en_US_vctk_p228.wav differ diff --git a/style_vector/en_US_vctk_p229.wav b/style_vector/en_US_vctk_p229.wav new file mode 100644 index 0000000000000000000000000000000000000000..52de5628fbc22dbfd1fb459e8b39ddd02657b7df Binary files /dev/null and b/style_vector/en_US_vctk_p229.wav differ diff --git a/style_vector/en_US_vctk_p230.wav b/style_vector/en_US_vctk_p230.wav new file mode 100644 index 0000000000000000000000000000000000000000..ce118a9ef072ee607aeaacdd3df87a4f329687d4 Binary files /dev/null and b/style_vector/en_US_vctk_p230.wav differ diff --git a/style_vector/en_US_vctk_p231.wav b/style_vector/en_US_vctk_p231.wav new file mode 100644 index 0000000000000000000000000000000000000000..b1f7f016f684e9492d29138a827fa8c777897ac9 Binary files /dev/null and b/style_vector/en_US_vctk_p231.wav differ diff --git a/style_vector/en_US_vctk_p232.wav b/style_vector/en_US_vctk_p232.wav new file mode 100644 index 0000000000000000000000000000000000000000..622bacab68f5899a967ba632e07ead6d7ba871d6 Binary files /dev/null and b/style_vector/en_US_vctk_p232.wav differ diff --git a/style_vector/en_US_vctk_p233.wav b/style_vector/en_US_vctk_p233.wav new file mode 100644 index 0000000000000000000000000000000000000000..9696a13f8ed0e273918ef5996a9fe5240a52da6d Binary files /dev/null and b/style_vector/en_US_vctk_p233.wav differ diff --git a/style_vector/en_US_vctk_p234.wav b/style_vector/en_US_vctk_p234.wav new file mode 100644 index 0000000000000000000000000000000000000000..100fa7802b8a2138bbd97ab1064aa90c8bbd8ba1 Binary files /dev/null and b/style_vector/en_US_vctk_p234.wav differ diff --git a/style_vector/en_US_vctk_p236.wav b/style_vector/en_US_vctk_p236.wav new file mode 100644 index 0000000000000000000000000000000000000000..bc65864e23e4e5302f64f5676be07750d5058240 Binary files /dev/null and b/style_vector/en_US_vctk_p236.wav differ diff --git a/style_vector/en_US_vctk_p237.wav b/style_vector/en_US_vctk_p237.wav new file mode 100644 index 0000000000000000000000000000000000000000..58b75a0442b66b1148e4fb45dd82efe9b685f821 Binary files /dev/null and b/style_vector/en_US_vctk_p237.wav differ diff --git a/style_vector/en_US_vctk_p238.wav b/style_vector/en_US_vctk_p238.wav new file mode 100644 index 0000000000000000000000000000000000000000..c50f727ffe8e4167cdd792e86b9846cc9d37837b Binary files /dev/null and b/style_vector/en_US_vctk_p238.wav differ diff --git a/style_vector/en_US_vctk_p239.wav b/style_vector/en_US_vctk_p239.wav new file mode 100644 index 0000000000000000000000000000000000000000..600addb31de3ee0ba6d18ff4c7c1dda4dfc5eb36 Binary files /dev/null and b/style_vector/en_US_vctk_p239.wav differ diff --git a/style_vector/en_US_vctk_p240.wav b/style_vector/en_US_vctk_p240.wav new file mode 100644 index 0000000000000000000000000000000000000000..23f54c271a99f84a28b7e3a111315e9216a2a495 Binary files /dev/null and b/style_vector/en_US_vctk_p240.wav differ diff --git a/style_vector/en_US_vctk_p241.wav b/style_vector/en_US_vctk_p241.wav new file mode 100644 index 0000000000000000000000000000000000000000..1765f988c211683d2d84dd5b88c3a95e5a33bade Binary files /dev/null and b/style_vector/en_US_vctk_p241.wav differ diff --git a/style_vector/en_US_vctk_p243.wav b/style_vector/en_US_vctk_p243.wav new file mode 100644 index 0000000000000000000000000000000000000000..c65e51a57ad285bb20d331c057da65cdffe76847 Binary files /dev/null and b/style_vector/en_US_vctk_p243.wav differ diff --git a/style_vector/en_US_vctk_p244.wav b/style_vector/en_US_vctk_p244.wav new file mode 100644 index 0000000000000000000000000000000000000000..a1bd15057a64d02aceb71e065ada815bef9b1fe6 Binary files /dev/null and b/style_vector/en_US_vctk_p244.wav differ diff --git a/style_vector/en_US_vctk_p245.wav b/style_vector/en_US_vctk_p245.wav new file mode 100644 index 0000000000000000000000000000000000000000..e6739f6dd6212e05bfbf9296f6131e2d08d294b2 Binary files /dev/null and b/style_vector/en_US_vctk_p245.wav differ diff --git a/style_vector/en_US_vctk_p246.wav b/style_vector/en_US_vctk_p246.wav new file mode 100644 index 0000000000000000000000000000000000000000..3d20a50a80fc854bf3a1f3babfd0969d7a7989d4 Binary files /dev/null and b/style_vector/en_US_vctk_p246.wav differ diff --git a/style_vector/en_US_vctk_p247.wav b/style_vector/en_US_vctk_p247.wav new file mode 100644 index 0000000000000000000000000000000000000000..b5988ea4aa30917d8d35060449e1c2e2fa995e2d Binary files /dev/null and b/style_vector/en_US_vctk_p247.wav differ diff --git a/style_vector/en_US_vctk_p248.wav b/style_vector/en_US_vctk_p248.wav new file mode 100644 index 0000000000000000000000000000000000000000..5c6a229ee0c52a93e12528c3b852939ce1328b21 Binary files /dev/null and b/style_vector/en_US_vctk_p248.wav differ diff --git a/style_vector/en_US_vctk_p249.wav b/style_vector/en_US_vctk_p249.wav new file mode 100644 index 0000000000000000000000000000000000000000..8ff3e0fa55f11d0d10b7b6136f2c0704691739dc Binary files /dev/null and b/style_vector/en_US_vctk_p249.wav differ diff --git a/style_vector/en_US_vctk_p250.wav b/style_vector/en_US_vctk_p250.wav new file mode 100644 index 0000000000000000000000000000000000000000..cfdc103366fcf51c3d2ce5e99d1b63acab4ed617 Binary files /dev/null and b/style_vector/en_US_vctk_p250.wav differ diff --git a/style_vector/en_US_vctk_p251.wav b/style_vector/en_US_vctk_p251.wav new file mode 100644 index 0000000000000000000000000000000000000000..f411c3e72ff0ddc1a66c7462b333169acf40f2fc Binary files /dev/null and b/style_vector/en_US_vctk_p251.wav differ diff --git a/style_vector/en_US_vctk_p252.wav b/style_vector/en_US_vctk_p252.wav new file mode 100644 index 0000000000000000000000000000000000000000..8e5f24391ca23249ca48b9953a3ab662a69334e3 Binary files /dev/null and b/style_vector/en_US_vctk_p252.wav differ diff --git a/style_vector/en_US_vctk_p253.wav b/style_vector/en_US_vctk_p253.wav new file mode 100644 index 0000000000000000000000000000000000000000..8eacab25181e7afd97bfcc6eed21e633a2a700b0 Binary files /dev/null and b/style_vector/en_US_vctk_p253.wav differ diff --git a/style_vector/en_US_vctk_p254.wav b/style_vector/en_US_vctk_p254.wav new file mode 100644 index 0000000000000000000000000000000000000000..e9ae1d5ec376a225cabce53092d905d2debf9bcf Binary files /dev/null and b/style_vector/en_US_vctk_p254.wav differ diff --git a/style_vector/en_US_vctk_p255.wav b/style_vector/en_US_vctk_p255.wav new file mode 100644 index 0000000000000000000000000000000000000000..c936b5fc77752063659cba540da28c186e5ff866 Binary files /dev/null and b/style_vector/en_US_vctk_p255.wav differ diff --git a/style_vector/en_US_vctk_p256.wav b/style_vector/en_US_vctk_p256.wav new file mode 100644 index 0000000000000000000000000000000000000000..a616a191bfa4de78cf5a6520766e3f5f0b91a693 Binary files /dev/null and b/style_vector/en_US_vctk_p256.wav differ diff --git a/style_vector/en_US_vctk_p257.wav b/style_vector/en_US_vctk_p257.wav new file mode 100644 index 0000000000000000000000000000000000000000..b2ebe4d4ccc8e0f0eb9d54a67504718df8a02b80 Binary files /dev/null and b/style_vector/en_US_vctk_p257.wav differ diff --git a/style_vector/en_US_vctk_p258.wav b/style_vector/en_US_vctk_p258.wav new file mode 100644 index 0000000000000000000000000000000000000000..49b2ada769a6109ad13106c6a3a3bd971b940f79 Binary files /dev/null and b/style_vector/en_US_vctk_p258.wav differ diff --git a/style_vector/en_US_vctk_p259.wav b/style_vector/en_US_vctk_p259.wav new file mode 100644 index 0000000000000000000000000000000000000000..11f71f84d2b3f919ecae2dbf51df1967ac1098b6 Binary files /dev/null and b/style_vector/en_US_vctk_p259.wav differ diff --git a/style_vector/en_US_vctk_p260.wav b/style_vector/en_US_vctk_p260.wav new file mode 100644 index 0000000000000000000000000000000000000000..43177e807cf13f2a58735ffcee7e1592991d62af Binary files /dev/null and b/style_vector/en_US_vctk_p260.wav differ diff --git a/style_vector/en_US_vctk_p261.wav b/style_vector/en_US_vctk_p261.wav new file mode 100644 index 0000000000000000000000000000000000000000..6c90913e4dff3bc31e4ff8ee48176be4aef701c3 Binary files /dev/null and b/style_vector/en_US_vctk_p261.wav differ diff --git a/style_vector/en_US_vctk_p262.wav b/style_vector/en_US_vctk_p262.wav new file mode 100644 index 0000000000000000000000000000000000000000..c6fbef9cf0775a7425af3143e4e46f3278c04d31 Binary files /dev/null and b/style_vector/en_US_vctk_p262.wav differ diff --git a/style_vector/en_US_vctk_p263.wav b/style_vector/en_US_vctk_p263.wav new file mode 100644 index 0000000000000000000000000000000000000000..ec34643012a5621e71446f4d7dd5cc8e3ae61013 Binary files /dev/null and b/style_vector/en_US_vctk_p263.wav differ diff --git a/style_vector/en_US_vctk_p264.wav b/style_vector/en_US_vctk_p264.wav new file mode 100644 index 0000000000000000000000000000000000000000..97358500759adceedeb5769f3e84de0695ec118f Binary files /dev/null and b/style_vector/en_US_vctk_p264.wav differ diff --git a/style_vector/en_US_vctk_p265.wav b/style_vector/en_US_vctk_p265.wav new file mode 100644 index 0000000000000000000000000000000000000000..7f62a976524aac32602ce7e209548d8f9bd5e5db Binary files /dev/null and b/style_vector/en_US_vctk_p265.wav differ diff --git a/style_vector/en_US_vctk_p266.wav b/style_vector/en_US_vctk_p266.wav new file mode 100644 index 0000000000000000000000000000000000000000..9d81df351d0d977ae7a44323b57eae9b0733364f Binary files /dev/null and b/style_vector/en_US_vctk_p266.wav differ diff --git a/style_vector/en_US_vctk_p267.wav b/style_vector/en_US_vctk_p267.wav new file mode 100644 index 0000000000000000000000000000000000000000..4a7b4505a60e6e2785277d241fff87caf9ba08b3 Binary files /dev/null and b/style_vector/en_US_vctk_p267.wav differ diff --git a/style_vector/en_US_vctk_p268.wav b/style_vector/en_US_vctk_p268.wav new file mode 100644 index 0000000000000000000000000000000000000000..38b152007b9f10f4ea97d6c0b4613176aae9a7be Binary files /dev/null and b/style_vector/en_US_vctk_p268.wav differ diff --git a/style_vector/en_US_vctk_p269.wav b/style_vector/en_US_vctk_p269.wav new file mode 100644 index 0000000000000000000000000000000000000000..a428ad6f4c0b15c30759203f053ecec50d32d23b Binary files /dev/null and b/style_vector/en_US_vctk_p269.wav differ diff --git a/style_vector/en_US_vctk_p270.wav b/style_vector/en_US_vctk_p270.wav new file mode 100644 index 0000000000000000000000000000000000000000..e981996a54bfb317908504f14686f1d56bb082fe Binary files /dev/null and b/style_vector/en_US_vctk_p270.wav differ diff --git a/style_vector/en_US_vctk_p271.wav b/style_vector/en_US_vctk_p271.wav new file mode 100644 index 0000000000000000000000000000000000000000..13428a593ef6be5f429e016ee6b966a4a8bccf34 Binary files /dev/null and b/style_vector/en_US_vctk_p271.wav differ diff --git a/style_vector/en_US_vctk_p272.wav b/style_vector/en_US_vctk_p272.wav new file mode 100644 index 0000000000000000000000000000000000000000..01c67c7ef2d2c357565c2715b11ba02a8520ea59 Binary files /dev/null and b/style_vector/en_US_vctk_p272.wav differ diff --git a/style_vector/en_US_vctk_p273.wav b/style_vector/en_US_vctk_p273.wav new file mode 100644 index 0000000000000000000000000000000000000000..2adb85c0b86a6b0a07a9f441544309e696d9bd48 Binary files /dev/null and b/style_vector/en_US_vctk_p273.wav differ diff --git a/style_vector/en_US_vctk_p274.wav b/style_vector/en_US_vctk_p274.wav new file mode 100644 index 0000000000000000000000000000000000000000..bd05e62eec0ff4ee020945a01f0e44de7d7adccc Binary files /dev/null and b/style_vector/en_US_vctk_p274.wav differ diff --git a/style_vector/en_US_vctk_p275.wav b/style_vector/en_US_vctk_p275.wav new file mode 100644 index 0000000000000000000000000000000000000000..ffae1f367f0c075e479987500cff5c184ca6d00e Binary files /dev/null and b/style_vector/en_US_vctk_p275.wav differ diff --git a/style_vector/en_US_vctk_p276.wav b/style_vector/en_US_vctk_p276.wav new file mode 100644 index 0000000000000000000000000000000000000000..49b65a7425b0712bbe71bdb2b9f08ce63cfd8633 Binary files /dev/null and b/style_vector/en_US_vctk_p276.wav differ diff --git a/style_vector/en_US_vctk_p277.wav b/style_vector/en_US_vctk_p277.wav new file mode 100644 index 0000000000000000000000000000000000000000..287e4501c7d849fe071ed81f0983503beee20770 Binary files /dev/null and b/style_vector/en_US_vctk_p277.wav differ diff --git a/style_vector/en_US_vctk_p278.wav b/style_vector/en_US_vctk_p278.wav new file mode 100644 index 0000000000000000000000000000000000000000..f4a25c7f5e9bbe11b11b457e24f62fe3645f7cbc Binary files /dev/null and b/style_vector/en_US_vctk_p278.wav differ diff --git a/style_vector/en_US_vctk_p279.wav b/style_vector/en_US_vctk_p279.wav new file mode 100644 index 0000000000000000000000000000000000000000..e2d3efff41d4de163f11efac929c9fd5dce38406 Binary files /dev/null and b/style_vector/en_US_vctk_p279.wav differ diff --git a/style_vector/en_US_vctk_p280.wav b/style_vector/en_US_vctk_p280.wav new file mode 100644 index 0000000000000000000000000000000000000000..45d80b25d0061f7e4464ce26c7e439efc864bae2 Binary files /dev/null and b/style_vector/en_US_vctk_p280.wav differ diff --git a/style_vector/en_US_vctk_p281.wav b/style_vector/en_US_vctk_p281.wav new file mode 100644 index 0000000000000000000000000000000000000000..e3073316792a19c44cd34f368ba7b4f54e8c6649 Binary files /dev/null and b/style_vector/en_US_vctk_p281.wav differ diff --git a/style_vector/en_US_vctk_p282.wav b/style_vector/en_US_vctk_p282.wav new file mode 100644 index 0000000000000000000000000000000000000000..103619125acacc06875f05370a457b630431885d Binary files /dev/null and b/style_vector/en_US_vctk_p282.wav differ diff --git a/style_vector/en_US_vctk_p283.wav b/style_vector/en_US_vctk_p283.wav new file mode 100644 index 0000000000000000000000000000000000000000..8496ee38819513c0d8ad65cdd957709901f990c3 Binary files /dev/null and b/style_vector/en_US_vctk_p283.wav differ diff --git a/style_vector/en_US_vctk_p284.wav b/style_vector/en_US_vctk_p284.wav new file mode 100644 index 0000000000000000000000000000000000000000..864693eaba7ee00c19953999bc4a0bac66f8b270 Binary files /dev/null and b/style_vector/en_US_vctk_p284.wav differ diff --git a/style_vector/en_US_vctk_p285.wav b/style_vector/en_US_vctk_p285.wav new file mode 100644 index 0000000000000000000000000000000000000000..ea0e759922eaf6d1240fd10c94656fa91fd36b8a Binary files /dev/null and b/style_vector/en_US_vctk_p285.wav differ diff --git a/style_vector/en_US_vctk_p286.wav b/style_vector/en_US_vctk_p286.wav new file mode 100644 index 0000000000000000000000000000000000000000..b726bd234c5318f0978cb9669cb993a218e54bac Binary files /dev/null and b/style_vector/en_US_vctk_p286.wav differ diff --git a/style_vector/en_US_vctk_p287.wav b/style_vector/en_US_vctk_p287.wav new file mode 100644 index 0000000000000000000000000000000000000000..224ce96d3530c9f4146cb5b6f3d0295738973b9d Binary files /dev/null and b/style_vector/en_US_vctk_p287.wav differ diff --git a/style_vector/en_US_vctk_p288.wav b/style_vector/en_US_vctk_p288.wav new file mode 100644 index 0000000000000000000000000000000000000000..bf9c70c46249b95f9ab175b45d7c973aee98979e Binary files /dev/null and b/style_vector/en_US_vctk_p288.wav differ diff --git a/style_vector/en_US_vctk_p292.wav b/style_vector/en_US_vctk_p292.wav new file mode 100644 index 0000000000000000000000000000000000000000..1b7f734c2d6c9bf0eb153f5bdd4030a9df839eb3 Binary files /dev/null and b/style_vector/en_US_vctk_p292.wav differ diff --git a/style_vector/en_US_vctk_p293.wav b/style_vector/en_US_vctk_p293.wav new file mode 100644 index 0000000000000000000000000000000000000000..65826bc0af542ed0bef356a172e6b15c9cbac687 Binary files /dev/null and b/style_vector/en_US_vctk_p293.wav differ diff --git a/style_vector/en_US_vctk_p294.wav b/style_vector/en_US_vctk_p294.wav new file mode 100644 index 0000000000000000000000000000000000000000..16c462a7f6634f3d1944e2c98991b27e9569febc Binary files /dev/null and b/style_vector/en_US_vctk_p294.wav differ diff --git a/style_vector/en_US_vctk_p295.wav b/style_vector/en_US_vctk_p295.wav new file mode 100644 index 0000000000000000000000000000000000000000..c6b32684c5440e92088e86b9eb745b2466999f18 Binary files /dev/null and b/style_vector/en_US_vctk_p295.wav differ diff --git a/style_vector/en_US_vctk_p297.wav b/style_vector/en_US_vctk_p297.wav new file mode 100644 index 0000000000000000000000000000000000000000..81b54d0a9a4bb13b74071738c02803daeb7af6e6 Binary files /dev/null and b/style_vector/en_US_vctk_p297.wav differ diff --git a/style_vector/en_US_vctk_p298.wav b/style_vector/en_US_vctk_p298.wav new file mode 100644 index 0000000000000000000000000000000000000000..9f94a481f5f7b78718e8f3f213eda90ccac7731d Binary files /dev/null and b/style_vector/en_US_vctk_p298.wav differ diff --git a/style_vector/en_US_vctk_p299.wav b/style_vector/en_US_vctk_p299.wav new file mode 100644 index 0000000000000000000000000000000000000000..cb522b9b5cad03f66886a94738ef84ec157cb569 Binary files /dev/null and b/style_vector/en_US_vctk_p299.wav differ diff --git a/style_vector/en_US_vctk_p300.wav b/style_vector/en_US_vctk_p300.wav new file mode 100644 index 0000000000000000000000000000000000000000..7bf944bc07e8e1f9b008a6fa70557503129dd5f9 Binary files /dev/null and b/style_vector/en_US_vctk_p300.wav differ diff --git a/style_vector/en_US_vctk_p301.wav b/style_vector/en_US_vctk_p301.wav new file mode 100644 index 0000000000000000000000000000000000000000..f2efc5aeff4808cf2e4cd2a250c5cdd0deca969d Binary files /dev/null and b/style_vector/en_US_vctk_p301.wav differ diff --git a/style_vector/en_US_vctk_p302.wav b/style_vector/en_US_vctk_p302.wav new file mode 100644 index 0000000000000000000000000000000000000000..dc67a9494eb25837e388219a60e66eae5a7509a8 Binary files /dev/null and b/style_vector/en_US_vctk_p302.wav differ diff --git a/style_vector/en_US_vctk_p303.wav b/style_vector/en_US_vctk_p303.wav new file mode 100644 index 0000000000000000000000000000000000000000..5766fe0d66c625b031ed1d453bee2517fb66d434 Binary files /dev/null and b/style_vector/en_US_vctk_p303.wav differ diff --git a/style_vector/en_US_vctk_p304.wav b/style_vector/en_US_vctk_p304.wav new file mode 100644 index 0000000000000000000000000000000000000000..59f42eb5ba2e371b132bb53828f05f70817e5f0f Binary files /dev/null and b/style_vector/en_US_vctk_p304.wav differ diff --git a/style_vector/en_US_vctk_p305.wav b/style_vector/en_US_vctk_p305.wav new file mode 100644 index 0000000000000000000000000000000000000000..21711e3a8e9f786107c85a39f5d8612d69af8602 Binary files /dev/null and b/style_vector/en_US_vctk_p305.wav differ diff --git a/style_vector/en_US_vctk_p306.wav b/style_vector/en_US_vctk_p306.wav new file mode 100644 index 0000000000000000000000000000000000000000..603664bd84a1fbfea4cd2843600512ac8f00bdd7 Binary files /dev/null and b/style_vector/en_US_vctk_p306.wav differ diff --git a/style_vector/en_US_vctk_p307.wav b/style_vector/en_US_vctk_p307.wav new file mode 100644 index 0000000000000000000000000000000000000000..dadfe865533a20071cd9286b5f3ae34b7b80bedb Binary files /dev/null and b/style_vector/en_US_vctk_p307.wav differ diff --git a/style_vector/en_US_vctk_p308.wav b/style_vector/en_US_vctk_p308.wav new file mode 100644 index 0000000000000000000000000000000000000000..c441c7382a48307b4fa1a44cab78ff39ebba9d36 Binary files /dev/null and b/style_vector/en_US_vctk_p308.wav differ diff --git a/style_vector/en_US_vctk_p310.wav b/style_vector/en_US_vctk_p310.wav new file mode 100644 index 0000000000000000000000000000000000000000..80c9b0d3cb94b6b9228c77e40367fb8da6be2319 Binary files /dev/null and b/style_vector/en_US_vctk_p310.wav differ diff --git a/style_vector/en_US_vctk_p311.wav b/style_vector/en_US_vctk_p311.wav new file mode 100644 index 0000000000000000000000000000000000000000..44e125e8896d6d97d9ab471481f8661530206929 Binary files /dev/null and b/style_vector/en_US_vctk_p311.wav differ diff --git a/style_vector/en_US_vctk_p312.wav b/style_vector/en_US_vctk_p312.wav new file mode 100644 index 0000000000000000000000000000000000000000..f37e7a89cf23c5f89071a919fa240949a6d25a46 Binary files /dev/null and b/style_vector/en_US_vctk_p312.wav differ diff --git a/style_vector/en_US_vctk_p313.wav b/style_vector/en_US_vctk_p313.wav new file mode 100644 index 0000000000000000000000000000000000000000..06dcd393378d30f513291fba57d73c52c51a83c8 Binary files /dev/null and b/style_vector/en_US_vctk_p313.wav differ diff --git a/style_vector/en_US_vctk_p314.wav b/style_vector/en_US_vctk_p314.wav new file mode 100644 index 0000000000000000000000000000000000000000..4a5eec7ab270ec34c9b21f816824972ff6e25384 Binary files /dev/null and b/style_vector/en_US_vctk_p314.wav differ diff --git a/style_vector/en_US_vctk_p316.wav b/style_vector/en_US_vctk_p316.wav new file mode 100644 index 0000000000000000000000000000000000000000..99e6b90f56bca3ea31dc87144a3baad7d2a7191c Binary files /dev/null and b/style_vector/en_US_vctk_p316.wav differ diff --git a/style_vector/en_US_vctk_p317.wav b/style_vector/en_US_vctk_p317.wav new file mode 100644 index 0000000000000000000000000000000000000000..6e82cd350d5704500d7955ca492b6d0eddfc7a10 Binary files /dev/null and b/style_vector/en_US_vctk_p317.wav differ diff --git a/style_vector/en_US_vctk_p318.wav b/style_vector/en_US_vctk_p318.wav new file mode 100644 index 0000000000000000000000000000000000000000..18f8b9196c0c15d0d58a8f7c6a0d74d286c75eb6 Binary files /dev/null and b/style_vector/en_US_vctk_p318.wav differ diff --git a/style_vector/en_US_vctk_p323.wav b/style_vector/en_US_vctk_p323.wav new file mode 100644 index 0000000000000000000000000000000000000000..214c617f6e640708e9f6960c687b833d42549392 Binary files /dev/null and b/style_vector/en_US_vctk_p323.wav differ diff --git a/style_vector/en_US_vctk_p326.wav b/style_vector/en_US_vctk_p326.wav new file mode 100644 index 0000000000000000000000000000000000000000..01bf9db1336e5db425ca986b837e1f3ec6662ea3 Binary files /dev/null and b/style_vector/en_US_vctk_p326.wav differ diff --git a/style_vector/en_US_vctk_p329.wav b/style_vector/en_US_vctk_p329.wav new file mode 100644 index 0000000000000000000000000000000000000000..8fd021452efea1fefb63894f7ceb8b491e71c0b7 Binary files /dev/null and b/style_vector/en_US_vctk_p329.wav differ diff --git a/style_vector/en_US_vctk_p330.wav b/style_vector/en_US_vctk_p330.wav new file mode 100644 index 0000000000000000000000000000000000000000..04b75ce1a5dee7cbb117ee4c709df44913360bb1 Binary files /dev/null and b/style_vector/en_US_vctk_p330.wav differ diff --git a/style_vector/en_US_vctk_p333.wav b/style_vector/en_US_vctk_p333.wav new file mode 100644 index 0000000000000000000000000000000000000000..5bd0f096f5e777ef1afb62b7903e7903dbb7a4f9 Binary files /dev/null and b/style_vector/en_US_vctk_p333.wav differ diff --git a/style_vector/en_US_vctk_p334.wav b/style_vector/en_US_vctk_p334.wav new file mode 100644 index 0000000000000000000000000000000000000000..f6d545d6b9a183bb3572b1a38aaa966054ef55a9 Binary files /dev/null and b/style_vector/en_US_vctk_p334.wav differ diff --git a/style_vector/en_US_vctk_p335.wav b/style_vector/en_US_vctk_p335.wav new file mode 100644 index 0000000000000000000000000000000000000000..4948089b36a4afb8eba50964ecb3de88f33151c6 Binary files /dev/null and b/style_vector/en_US_vctk_p335.wav differ diff --git a/style_vector/en_US_vctk_p336.wav b/style_vector/en_US_vctk_p336.wav new file mode 100644 index 0000000000000000000000000000000000000000..01a19a580adf4ef31340af8eadc330b5ed55c4cc Binary files /dev/null and b/style_vector/en_US_vctk_p336.wav differ diff --git a/style_vector/en_US_vctk_p339.wav b/style_vector/en_US_vctk_p339.wav new file mode 100644 index 0000000000000000000000000000000000000000..ea8165286561a79e9c3ab95032226b16caadd6b1 Binary files /dev/null and b/style_vector/en_US_vctk_p339.wav differ diff --git a/style_vector/en_US_vctk_p340.wav b/style_vector/en_US_vctk_p340.wav new file mode 100644 index 0000000000000000000000000000000000000000..6a06c795a09a891c1cd728d0ddf2506666794f76 Binary files /dev/null and b/style_vector/en_US_vctk_p340.wav differ diff --git a/style_vector/en_US_vctk_p341.wav b/style_vector/en_US_vctk_p341.wav new file mode 100644 index 0000000000000000000000000000000000000000..47d45cbd1905290f694d8d634493d0ed851e5242 Binary files /dev/null and b/style_vector/en_US_vctk_p341.wav differ diff --git a/style_vector/en_US_vctk_p343.wav b/style_vector/en_US_vctk_p343.wav new file mode 100644 index 0000000000000000000000000000000000000000..78bcd0f069b36715e8513f345964f7a73979cd02 Binary files /dev/null and b/style_vector/en_US_vctk_p343.wav differ diff --git a/style_vector/en_US_vctk_p345.wav b/style_vector/en_US_vctk_p345.wav new file mode 100644 index 0000000000000000000000000000000000000000..1c3f6bad11c3152ed71a560f7ed732ea68571f0f Binary files /dev/null and b/style_vector/en_US_vctk_p345.wav differ diff --git a/style_vector/en_US_vctk_p347.wav b/style_vector/en_US_vctk_p347.wav new file mode 100644 index 0000000000000000000000000000000000000000..e1943ea81fbb7e9daac7f6073df7e7f73026869a Binary files /dev/null and b/style_vector/en_US_vctk_p347.wav differ diff --git a/style_vector/en_US_vctk_p351.wav b/style_vector/en_US_vctk_p351.wav new file mode 100644 index 0000000000000000000000000000000000000000..b5b896039e6a91da5ea60308d6b7a2f1df58e675 Binary files /dev/null and b/style_vector/en_US_vctk_p351.wav differ diff --git a/style_vector/en_US_vctk_p360.wav b/style_vector/en_US_vctk_p360.wav new file mode 100644 index 0000000000000000000000000000000000000000..58c2b9632de92cbc216656cd6825cc3c2aa9d94f Binary files /dev/null and b/style_vector/en_US_vctk_p360.wav differ diff --git a/style_vector/en_US_vctk_p361.wav b/style_vector/en_US_vctk_p361.wav new file mode 100644 index 0000000000000000000000000000000000000000..8bae5ad07a9f74a35fa64fa482e416ae606b9b89 Binary files /dev/null and b/style_vector/en_US_vctk_p361.wav differ diff --git a/style_vector/en_US_vctk_p362.wav b/style_vector/en_US_vctk_p362.wav new file mode 100644 index 0000000000000000000000000000000000000000..54bd4435edb9b1cf6f057c7b5fd49f7172ed5197 Binary files /dev/null and b/style_vector/en_US_vctk_p362.wav differ diff --git a/style_vector/en_US_vctk_p363.wav b/style_vector/en_US_vctk_p363.wav new file mode 100644 index 0000000000000000000000000000000000000000..c6e044cbbc5b0755dbc7e4f5ab0fb0353a1f4b9d Binary files /dev/null and b/style_vector/en_US_vctk_p363.wav differ diff --git a/style_vector/en_US_vctk_p364.wav b/style_vector/en_US_vctk_p364.wav new file mode 100644 index 0000000000000000000000000000000000000000..d1af4b674c3a0b9c17aa3ce462d4d1921f7755c3 Binary files /dev/null and b/style_vector/en_US_vctk_p364.wav differ diff --git a/style_vector/en_US_vctk_p374.wav b/style_vector/en_US_vctk_p374.wav new file mode 100644 index 0000000000000000000000000000000000000000..6db1268facc41f45177aaa8adaf8a2e672f06068 Binary files /dev/null and b/style_vector/en_US_vctk_p374.wav differ diff --git a/style_vector/en_US_vctk_p376.wav b/style_vector/en_US_vctk_p376.wav new file mode 100644 index 0000000000000000000000000000000000000000..2223eba36dadf9dc3cd2734c1264d8436e58fb99 Binary files /dev/null and b/style_vector/en_US_vctk_p376.wav differ diff --git a/style_vector/en_US_vctk_s5.wav b/style_vector/en_US_vctk_s5.wav new file mode 100644 index 0000000000000000000000000000000000000000..40909b6654e1bccd2629f5bd5c5840d943487a59 Binary files /dev/null and b/style_vector/en_US_vctk_s5.wav differ diff --git a/style_vector_v2/en_UK_apope.wav b/style_vector_v2/en_UK_apope.wav new file mode 100644 index 0000000000000000000000000000000000000000..eb0842e2359cdd6c5932486c59bccec323b57d75 Binary files /dev/null and b/style_vector_v2/en_UK_apope.wav differ diff --git a/style_vector_v2/en_US_cmu_arctic_aew.wav b/style_vector_v2/en_US_cmu_arctic_aew.wav new file mode 100644 index 0000000000000000000000000000000000000000..765e5b092b2c14088d3a62e08cdea58dd35c1672 Binary files /dev/null and b/style_vector_v2/en_US_cmu_arctic_aew.wav differ diff --git a/style_vector_v2/en_US_cmu_arctic_ahw.wav b/style_vector_v2/en_US_cmu_arctic_ahw.wav new file mode 100644 index 0000000000000000000000000000000000000000..1bad647b8008c59a5b987742244fcdc5fcca1956 Binary files /dev/null and b/style_vector_v2/en_US_cmu_arctic_ahw.wav differ diff --git a/style_vector_v2/en_US_cmu_arctic_aup.wav b/style_vector_v2/en_US_cmu_arctic_aup.wav new file mode 100644 index 0000000000000000000000000000000000000000..d9456dc18559a3f02203b00969c57e5e9d693e4b Binary files /dev/null and b/style_vector_v2/en_US_cmu_arctic_aup.wav differ diff --git a/style_vector_v2/en_US_cmu_arctic_awbrms.wav b/style_vector_v2/en_US_cmu_arctic_awbrms.wav new file mode 100644 index 0000000000000000000000000000000000000000..e7ecd19105bab5b18537adc3b65efd7a1d41e7c1 Binary files /dev/null and b/style_vector_v2/en_US_cmu_arctic_awbrms.wav differ diff --git a/style_vector_v2/en_US_cmu_arctic_axb.wav b/style_vector_v2/en_US_cmu_arctic_axb.wav new file mode 100644 index 0000000000000000000000000000000000000000..cc6fc5618ebf2a21c29be652a0d7d71fbf61fc55 Binary files /dev/null and b/style_vector_v2/en_US_cmu_arctic_axb.wav differ diff --git a/style_vector_v2/en_US_cmu_arctic_bdl.wav b/style_vector_v2/en_US_cmu_arctic_bdl.wav new file mode 100644 index 0000000000000000000000000000000000000000..1cf2817f67aa7fc021b3548e77dbfabb3e79c557 Binary files /dev/null and b/style_vector_v2/en_US_cmu_arctic_bdl.wav differ diff --git a/style_vector_v2/en_US_cmu_arctic_clb.wav b/style_vector_v2/en_US_cmu_arctic_clb.wav new file mode 100644 index 0000000000000000000000000000000000000000..18bcc0b6a04240f015fd9343b313b74202a1863e Binary files /dev/null and b/style_vector_v2/en_US_cmu_arctic_clb.wav differ diff --git a/style_vector_v2/en_US_cmu_arctic_eey.wav b/style_vector_v2/en_US_cmu_arctic_eey.wav new file mode 100644 index 0000000000000000000000000000000000000000..2f006fdf779a1b989829f568751eb628db245d70 Binary files /dev/null and b/style_vector_v2/en_US_cmu_arctic_eey.wav differ diff --git a/style_vector_v2/en_US_cmu_arctic_fem.wav b/style_vector_v2/en_US_cmu_arctic_fem.wav new file mode 100644 index 0000000000000000000000000000000000000000..ab7e8f239c809005f335bdeac3f7011e94462075 Binary files /dev/null and b/style_vector_v2/en_US_cmu_arctic_fem.wav differ diff --git a/style_vector_v2/en_US_cmu_arctic_gka.wav b/style_vector_v2/en_US_cmu_arctic_gka.wav new file mode 100644 index 0000000000000000000000000000000000000000..267aacd5af74105d5dbceeeaee98c523efd27b0b Binary files /dev/null and b/style_vector_v2/en_US_cmu_arctic_gka.wav differ diff --git a/style_vector_v2/en_US_cmu_arctic_jmk.wav b/style_vector_v2/en_US_cmu_arctic_jmk.wav new file mode 100644 index 0000000000000000000000000000000000000000..a7ed7b9913c7beebb3aa786e6c209db3b6e30abf Binary files /dev/null and b/style_vector_v2/en_US_cmu_arctic_jmk.wav differ diff --git a/style_vector_v2/en_US_cmu_arctic_ksp.wav b/style_vector_v2/en_US_cmu_arctic_ksp.wav new file mode 100644 index 0000000000000000000000000000000000000000..3a8ec8affc3e323b23cdbbdc80bffa0c84c9bfde Binary files /dev/null and b/style_vector_v2/en_US_cmu_arctic_ksp.wav differ diff --git a/style_vector_v2/en_US_cmu_arctic_ljm.wav b/style_vector_v2/en_US_cmu_arctic_ljm.wav new file mode 100644 index 0000000000000000000000000000000000000000..eb44dc78fd204db99be714113c9aeb3d4d97de7d Binary files /dev/null and b/style_vector_v2/en_US_cmu_arctic_ljm.wav differ diff --git a/style_vector_v2/en_US_cmu_arctic_lnh.wav b/style_vector_v2/en_US_cmu_arctic_lnh.wav new file mode 100644 index 0000000000000000000000000000000000000000..6980004a58a59b6aee0895d578fecb6fbd2b21f1 Binary files /dev/null and b/style_vector_v2/en_US_cmu_arctic_lnh.wav differ diff --git a/style_vector_v2/en_US_cmu_arctic_rxr.wav b/style_vector_v2/en_US_cmu_arctic_rxr.wav new file mode 100644 index 0000000000000000000000000000000000000000..45234fe610cb15df75b5b7ad387c5c1a1a24ee67 Binary files /dev/null and b/style_vector_v2/en_US_cmu_arctic_rxr.wav differ diff --git a/style_vector_v2/en_US_cmu_arctic_slp.wav b/style_vector_v2/en_US_cmu_arctic_slp.wav new file mode 100644 index 0000000000000000000000000000000000000000..4151cc333dc2a017fbd116d35568ac208182a859 Binary files /dev/null and b/style_vector_v2/en_US_cmu_arctic_slp.wav differ diff --git a/style_vector_v2/en_US_cmu_arctic_slt.wav b/style_vector_v2/en_US_cmu_arctic_slt.wav new file mode 100644 index 0000000000000000000000000000000000000000..9aa322dc20a410810b58e85b815c348b2bf5a8de Binary files /dev/null and b/style_vector_v2/en_US_cmu_arctic_slt.wav differ diff --git a/style_vector_v2/en_US_hifi-tts_6097.wav b/style_vector_v2/en_US_hifi-tts_6097.wav new file mode 100644 index 0000000000000000000000000000000000000000..b65e06b630d0a8a6d660c08dedf1ffb416179fe3 Binary files /dev/null and b/style_vector_v2/en_US_hifi-tts_6097.wav differ diff --git a/style_vector_v2/en_US_hifi-tts_9017.wav b/style_vector_v2/en_US_hifi-tts_9017.wav new file mode 100644 index 0000000000000000000000000000000000000000..2ca574b9d6e70e21c03d157f5ec0679ffd611063 Binary files /dev/null and b/style_vector_v2/en_US_hifi-tts_9017.wav differ diff --git a/style_vector_v2/en_US_hifi-tts_92.wav b/style_vector_v2/en_US_hifi-tts_92.wav new file mode 100644 index 0000000000000000000000000000000000000000..b461c9e1c414039710d35e715d0579c8f7a212ad Binary files /dev/null and b/style_vector_v2/en_US_hifi-tts_92.wav differ diff --git a/style_vector_v2/en_US_ljspeech.wav b/style_vector_v2/en_US_ljspeech.wav new file mode 100644 index 0000000000000000000000000000000000000000..c5d6f7d8a08c999e300bf873efc8db8d27a8842e Binary files /dev/null and b/style_vector_v2/en_US_ljspeech.wav differ diff --git a/style_vector_v2/en_US_m-ailabs_elliot_miller.wav b/style_vector_v2/en_US_m-ailabs_elliot_miller.wav new file mode 100644 index 0000000000000000000000000000000000000000..e8be46aa271329b37d86c4d012199084eac084bf Binary files /dev/null and b/style_vector_v2/en_US_m-ailabs_elliot_miller.wav differ diff --git a/style_vector_v2/en_US_m-ailabs_judy_bieber.wav b/style_vector_v2/en_US_m-ailabs_judy_bieber.wav new file mode 100644 index 0000000000000000000000000000000000000000..7510ad983af8ea1bfa35856e1893d1c445d099e6 Binary files /dev/null and b/style_vector_v2/en_US_m-ailabs_judy_bieber.wav differ diff --git a/style_vector_v2/en_US_m-ailabs_mary_ann.wav b/style_vector_v2/en_US_m-ailabs_mary_ann.wav new file mode 100644 index 0000000000000000000000000000000000000000..0059d10be9964404b0332601cb9fe8c0c84649fd Binary files /dev/null and b/style_vector_v2/en_US_m-ailabs_mary_ann.wav differ diff --git a/style_vector_v2/en_US_vctk_p225.wav b/style_vector_v2/en_US_vctk_p225.wav new file mode 100644 index 0000000000000000000000000000000000000000..3b5cf9a67072b9aa7ff1df9523afb22e0bd21297 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p225.wav differ diff --git a/style_vector_v2/en_US_vctk_p226.wav b/style_vector_v2/en_US_vctk_p226.wav new file mode 100644 index 0000000000000000000000000000000000000000..8c59216938b156d1b01a6edbc462957b6c76a0c8 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p226.wav differ diff --git a/style_vector_v2/en_US_vctk_p227.wav b/style_vector_v2/en_US_vctk_p227.wav new file mode 100644 index 0000000000000000000000000000000000000000..3ed1e3776708fa10ce07a362342cfab0ed56b3fb Binary files /dev/null and b/style_vector_v2/en_US_vctk_p227.wav differ diff --git a/style_vector_v2/en_US_vctk_p228.wav b/style_vector_v2/en_US_vctk_p228.wav new file mode 100644 index 0000000000000000000000000000000000000000..a6ed295620e12b7964528d4f11e0c3c166a09a60 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p228.wav differ diff --git a/style_vector_v2/en_US_vctk_p229.wav b/style_vector_v2/en_US_vctk_p229.wav new file mode 100644 index 0000000000000000000000000000000000000000..19b1f7ca0fa911b1021e7ff7b30bea96f498a919 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p229.wav differ diff --git a/style_vector_v2/en_US_vctk_p230.wav b/style_vector_v2/en_US_vctk_p230.wav new file mode 100644 index 0000000000000000000000000000000000000000..ea476b4ea4ff8e3b0722afe3a7357fb25184fe98 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p230.wav differ diff --git a/style_vector_v2/en_US_vctk_p231.wav b/style_vector_v2/en_US_vctk_p231.wav new file mode 100644 index 0000000000000000000000000000000000000000..57a3d2ce702744d3544bf51bf8b20228c727afb0 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p231.wav differ diff --git a/style_vector_v2/en_US_vctk_p232.wav b/style_vector_v2/en_US_vctk_p232.wav new file mode 100644 index 0000000000000000000000000000000000000000..e774e46a568cafefc214a7569a58b1aa4635b46c Binary files /dev/null and b/style_vector_v2/en_US_vctk_p232.wav differ diff --git a/style_vector_v2/en_US_vctk_p233.wav b/style_vector_v2/en_US_vctk_p233.wav new file mode 100644 index 0000000000000000000000000000000000000000..e9eac34066c23061123fc188ceaf62aebe931dc1 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p233.wav differ diff --git a/style_vector_v2/en_US_vctk_p234.wav b/style_vector_v2/en_US_vctk_p234.wav new file mode 100644 index 0000000000000000000000000000000000000000..833717856183f7ce23c67c9004c4cb8ede544be1 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p234.wav differ diff --git a/style_vector_v2/en_US_vctk_p236.wav b/style_vector_v2/en_US_vctk_p236.wav new file mode 100644 index 0000000000000000000000000000000000000000..6c236a84b7a823cd1b49924b80a8d697298b0ac0 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p236.wav differ diff --git a/style_vector_v2/en_US_vctk_p237.wav b/style_vector_v2/en_US_vctk_p237.wav new file mode 100644 index 0000000000000000000000000000000000000000..1f27076bfe645a8420a445ca280f215ca9840baf Binary files /dev/null and b/style_vector_v2/en_US_vctk_p237.wav differ diff --git a/style_vector_v2/en_US_vctk_p238.wav b/style_vector_v2/en_US_vctk_p238.wav new file mode 100644 index 0000000000000000000000000000000000000000..0733fae86cc26c0fde6d3f688b57e1d0abe8e1f0 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p238.wav differ diff --git a/style_vector_v2/en_US_vctk_p239.wav b/style_vector_v2/en_US_vctk_p239.wav new file mode 100644 index 0000000000000000000000000000000000000000..993d5ae0d6543c6ac55da8666ccc580a0ef2289f Binary files /dev/null and b/style_vector_v2/en_US_vctk_p239.wav differ diff --git a/style_vector_v2/en_US_vctk_p240.wav b/style_vector_v2/en_US_vctk_p240.wav new file mode 100644 index 0000000000000000000000000000000000000000..6bed36e90283d7c4a4e40b3814e1906ad75a3a7d Binary files /dev/null and b/style_vector_v2/en_US_vctk_p240.wav differ diff --git a/style_vector_v2/en_US_vctk_p241.wav b/style_vector_v2/en_US_vctk_p241.wav new file mode 100644 index 0000000000000000000000000000000000000000..7d5d48f0d69456e4cb82d93253a0c926f1690769 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p241.wav differ diff --git a/style_vector_v2/en_US_vctk_p243.wav b/style_vector_v2/en_US_vctk_p243.wav new file mode 100644 index 0000000000000000000000000000000000000000..0f5c2217e58c7e662664ef5dbb6647c72b332a98 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p243.wav differ diff --git a/style_vector_v2/en_US_vctk_p244.wav b/style_vector_v2/en_US_vctk_p244.wav new file mode 100644 index 0000000000000000000000000000000000000000..075e987372444c14c5de5ad0abfffbed3a43f985 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p244.wav differ diff --git a/style_vector_v2/en_US_vctk_p245.wav b/style_vector_v2/en_US_vctk_p245.wav new file mode 100644 index 0000000000000000000000000000000000000000..d02eb539cc1eaefeb8e5b185eebc7d46085c6132 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p245.wav differ diff --git a/style_vector_v2/en_US_vctk_p246.wav b/style_vector_v2/en_US_vctk_p246.wav new file mode 100644 index 0000000000000000000000000000000000000000..48f9d0133e73b36e2d72a117a3ae33f65444e90b Binary files /dev/null and b/style_vector_v2/en_US_vctk_p246.wav differ diff --git a/style_vector_v2/en_US_vctk_p247.wav b/style_vector_v2/en_US_vctk_p247.wav new file mode 100644 index 0000000000000000000000000000000000000000..1859291df4c33081d8e2d771c18a396c8e73b5fa Binary files /dev/null and b/style_vector_v2/en_US_vctk_p247.wav differ diff --git a/style_vector_v2/en_US_vctk_p248.wav b/style_vector_v2/en_US_vctk_p248.wav new file mode 100644 index 0000000000000000000000000000000000000000..e97f7e46259bc1830d7a7b606f6a9217df557178 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p248.wav differ diff --git a/style_vector_v2/en_US_vctk_p249.wav b/style_vector_v2/en_US_vctk_p249.wav new file mode 100644 index 0000000000000000000000000000000000000000..8c27ea8bbf2c0753d538ca45bfe84fb274eb3fad Binary files /dev/null and b/style_vector_v2/en_US_vctk_p249.wav differ diff --git a/style_vector_v2/en_US_vctk_p250.wav b/style_vector_v2/en_US_vctk_p250.wav new file mode 100644 index 0000000000000000000000000000000000000000..38e21c294df044e1c39f74ccc04942fb5654c2f7 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p250.wav differ diff --git a/style_vector_v2/en_US_vctk_p251.wav b/style_vector_v2/en_US_vctk_p251.wav new file mode 100644 index 0000000000000000000000000000000000000000..3bdffe0483895f546c613ad1b658aaf6a2c5d18b Binary files /dev/null and b/style_vector_v2/en_US_vctk_p251.wav differ diff --git a/style_vector_v2/en_US_vctk_p252.wav b/style_vector_v2/en_US_vctk_p252.wav new file mode 100644 index 0000000000000000000000000000000000000000..73db11aa58ade2f15740eb5fa4f837e06d31993b Binary files /dev/null and b/style_vector_v2/en_US_vctk_p252.wav differ diff --git a/style_vector_v2/en_US_vctk_p253.wav b/style_vector_v2/en_US_vctk_p253.wav new file mode 100644 index 0000000000000000000000000000000000000000..23726cc46cfe730c04488a54e7ae066aef47286c Binary files /dev/null and b/style_vector_v2/en_US_vctk_p253.wav differ diff --git a/style_vector_v2/en_US_vctk_p254.wav b/style_vector_v2/en_US_vctk_p254.wav new file mode 100644 index 0000000000000000000000000000000000000000..e8caebd045ab83064fa4613a83a916067080e93d Binary files /dev/null and b/style_vector_v2/en_US_vctk_p254.wav differ diff --git a/style_vector_v2/en_US_vctk_p255.wav b/style_vector_v2/en_US_vctk_p255.wav new file mode 100644 index 0000000000000000000000000000000000000000..0c6523810f80aefce918b71874ac785618a99bf7 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p255.wav differ diff --git a/style_vector_v2/en_US_vctk_p256.wav b/style_vector_v2/en_US_vctk_p256.wav new file mode 100644 index 0000000000000000000000000000000000000000..2cdbf8f1993960c11b93c5e5d8d243219cffaf88 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p256.wav differ diff --git a/style_vector_v2/en_US_vctk_p257.wav b/style_vector_v2/en_US_vctk_p257.wav new file mode 100644 index 0000000000000000000000000000000000000000..3f142febdbd6a32a064b8cdadcdd15846826e46a Binary files /dev/null and b/style_vector_v2/en_US_vctk_p257.wav differ diff --git a/style_vector_v2/en_US_vctk_p258.wav b/style_vector_v2/en_US_vctk_p258.wav new file mode 100644 index 0000000000000000000000000000000000000000..e11ddf71603b842c10202effdb22f649b66cc181 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p258.wav differ diff --git a/style_vector_v2/en_US_vctk_p259.wav b/style_vector_v2/en_US_vctk_p259.wav new file mode 100644 index 0000000000000000000000000000000000000000..adb990538429dfb59b889d474ceb3701608dbb74 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p259.wav differ diff --git a/style_vector_v2/en_US_vctk_p260.wav b/style_vector_v2/en_US_vctk_p260.wav new file mode 100644 index 0000000000000000000000000000000000000000..87539c5a09d22ead7affce4be920bbdd6cf229aa Binary files /dev/null and b/style_vector_v2/en_US_vctk_p260.wav differ diff --git a/style_vector_v2/en_US_vctk_p261.wav b/style_vector_v2/en_US_vctk_p261.wav new file mode 100644 index 0000000000000000000000000000000000000000..11daa9682e2fbc1ffc1f5eeb49fd1f5f307398f9 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p261.wav differ diff --git a/style_vector_v2/en_US_vctk_p262.wav b/style_vector_v2/en_US_vctk_p262.wav new file mode 100644 index 0000000000000000000000000000000000000000..3f7a84bb411b74c8f34ff1a2abf61ca305a0164d Binary files /dev/null and b/style_vector_v2/en_US_vctk_p262.wav differ diff --git a/style_vector_v2/en_US_vctk_p263.wav b/style_vector_v2/en_US_vctk_p263.wav new file mode 100644 index 0000000000000000000000000000000000000000..1a365593e777f178323a66b73043aaf65b265bb1 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p263.wav differ diff --git a/style_vector_v2/en_US_vctk_p264.wav b/style_vector_v2/en_US_vctk_p264.wav new file mode 100644 index 0000000000000000000000000000000000000000..7998ded11da3c3f09665c8cb8e2b2a18a30a9327 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p264.wav differ diff --git a/style_vector_v2/en_US_vctk_p265.wav b/style_vector_v2/en_US_vctk_p265.wav new file mode 100644 index 0000000000000000000000000000000000000000..e2f6db3112294b7a3acf120d54454a9be47bb7fe Binary files /dev/null and b/style_vector_v2/en_US_vctk_p265.wav differ diff --git a/style_vector_v2/en_US_vctk_p266.wav b/style_vector_v2/en_US_vctk_p266.wav new file mode 100644 index 0000000000000000000000000000000000000000..efaccc3d9353e75bcafef79c402a8824e1796a88 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p266.wav differ diff --git a/style_vector_v2/en_US_vctk_p267.wav b/style_vector_v2/en_US_vctk_p267.wav new file mode 100644 index 0000000000000000000000000000000000000000..4d057f984ea4252c2ae4f0963aab473b54b69745 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p267.wav differ diff --git a/style_vector_v2/en_US_vctk_p268.wav b/style_vector_v2/en_US_vctk_p268.wav new file mode 100644 index 0000000000000000000000000000000000000000..d85f7678c468f3ca993b29415be84e5df0e4e591 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p268.wav differ diff --git a/style_vector_v2/en_US_vctk_p269.wav b/style_vector_v2/en_US_vctk_p269.wav new file mode 100644 index 0000000000000000000000000000000000000000..958cfb566a1d61dfb56eee53970ddaebc2777a55 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p269.wav differ diff --git a/style_vector_v2/en_US_vctk_p270.wav b/style_vector_v2/en_US_vctk_p270.wav new file mode 100644 index 0000000000000000000000000000000000000000..e2b221369e5f8d9d3b7e3d730f45cb0d5dfe77e4 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p270.wav differ diff --git a/style_vector_v2/en_US_vctk_p271.wav b/style_vector_v2/en_US_vctk_p271.wav new file mode 100644 index 0000000000000000000000000000000000000000..896d22f98f5acfa153ea7836edeedee240427121 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p271.wav differ diff --git a/style_vector_v2/en_US_vctk_p272.wav b/style_vector_v2/en_US_vctk_p272.wav new file mode 100644 index 0000000000000000000000000000000000000000..72c4c7c9cac434c802769905b99b979d4fdaec50 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p272.wav differ diff --git a/style_vector_v2/en_US_vctk_p273.wav b/style_vector_v2/en_US_vctk_p273.wav new file mode 100644 index 0000000000000000000000000000000000000000..3e2f2a9216e29f9ac81fe6a6452703f348e8d4af Binary files /dev/null and b/style_vector_v2/en_US_vctk_p273.wav differ diff --git a/style_vector_v2/en_US_vctk_p274.wav b/style_vector_v2/en_US_vctk_p274.wav new file mode 100644 index 0000000000000000000000000000000000000000..8528a7b87b2c1609fa111c029f71df3785c257b6 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p274.wav differ diff --git a/style_vector_v2/en_US_vctk_p275.wav b/style_vector_v2/en_US_vctk_p275.wav new file mode 100644 index 0000000000000000000000000000000000000000..c027575f35fd28e80572c64969066ae1a8b3054b Binary files /dev/null and b/style_vector_v2/en_US_vctk_p275.wav differ diff --git a/style_vector_v2/en_US_vctk_p276.wav b/style_vector_v2/en_US_vctk_p276.wav new file mode 100644 index 0000000000000000000000000000000000000000..9687332d836dd574c3aa804d3c70b78996722691 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p276.wav differ diff --git a/style_vector_v2/en_US_vctk_p277.wav b/style_vector_v2/en_US_vctk_p277.wav new file mode 100644 index 0000000000000000000000000000000000000000..1e6e907142326423d4138168b45f202b1a5c039e Binary files /dev/null and b/style_vector_v2/en_US_vctk_p277.wav differ diff --git a/style_vector_v2/en_US_vctk_p278.wav b/style_vector_v2/en_US_vctk_p278.wav new file mode 100644 index 0000000000000000000000000000000000000000..580f2baed4006df5002917ad08bb5cdee109dba1 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p278.wav differ diff --git a/style_vector_v2/en_US_vctk_p279.wav b/style_vector_v2/en_US_vctk_p279.wav new file mode 100644 index 0000000000000000000000000000000000000000..aa6c7e4257e632a6c914de3729af8b9fadb9ab9f Binary files /dev/null and b/style_vector_v2/en_US_vctk_p279.wav differ diff --git a/style_vector_v2/en_US_vctk_p280.wav b/style_vector_v2/en_US_vctk_p280.wav new file mode 100644 index 0000000000000000000000000000000000000000..994b7c64036f79402844f7260222a817db2ee2ab Binary files /dev/null and b/style_vector_v2/en_US_vctk_p280.wav differ diff --git a/style_vector_v2/en_US_vctk_p281.wav b/style_vector_v2/en_US_vctk_p281.wav new file mode 100644 index 0000000000000000000000000000000000000000..6fe002fd997428398103d35d0a3b5d9754c80e47 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p281.wav differ diff --git a/style_vector_v2/en_US_vctk_p282.wav b/style_vector_v2/en_US_vctk_p282.wav new file mode 100644 index 0000000000000000000000000000000000000000..96af01bce29d535fce99a3b68956472dd16d08e8 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p282.wav differ diff --git a/style_vector_v2/en_US_vctk_p283.wav b/style_vector_v2/en_US_vctk_p283.wav new file mode 100644 index 0000000000000000000000000000000000000000..6356e13ef711137118a1f02ffbe4c530aec27bba Binary files /dev/null and b/style_vector_v2/en_US_vctk_p283.wav differ diff --git a/style_vector_v2/en_US_vctk_p284.wav b/style_vector_v2/en_US_vctk_p284.wav new file mode 100644 index 0000000000000000000000000000000000000000..c47004b7e1d0fdc975eef0b75d599a1001ddf2f7 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p284.wav differ diff --git a/style_vector_v2/en_US_vctk_p285.wav b/style_vector_v2/en_US_vctk_p285.wav new file mode 100644 index 0000000000000000000000000000000000000000..9ec4ef84e89594f689f529480c27f276f03e8e1a Binary files /dev/null and b/style_vector_v2/en_US_vctk_p285.wav differ diff --git a/style_vector_v2/en_US_vctk_p286.wav b/style_vector_v2/en_US_vctk_p286.wav new file mode 100644 index 0000000000000000000000000000000000000000..1cd833643edf472820c31350c811ba3f1c370420 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p286.wav differ diff --git a/style_vector_v2/en_US_vctk_p287.wav b/style_vector_v2/en_US_vctk_p287.wav new file mode 100644 index 0000000000000000000000000000000000000000..789714f27fe9af36a62e220bda9b7e8287d997b1 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p287.wav differ diff --git a/style_vector_v2/en_US_vctk_p288.wav b/style_vector_v2/en_US_vctk_p288.wav new file mode 100644 index 0000000000000000000000000000000000000000..f6af84200e7d2b5cc2ecfbbd64ac2870e5a3c515 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p288.wav differ diff --git a/style_vector_v2/en_US_vctk_p292.wav b/style_vector_v2/en_US_vctk_p292.wav new file mode 100644 index 0000000000000000000000000000000000000000..dc7d64833cf31d0fa11ecb89124efffaf343e326 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p292.wav differ diff --git a/style_vector_v2/en_US_vctk_p293.wav b/style_vector_v2/en_US_vctk_p293.wav new file mode 100644 index 0000000000000000000000000000000000000000..6573837ef3bf71966a9c85b117c8590e83d70db4 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p293.wav differ diff --git a/style_vector_v2/en_US_vctk_p294.wav b/style_vector_v2/en_US_vctk_p294.wav new file mode 100644 index 0000000000000000000000000000000000000000..25762b99ba33cd6bbab56f687bb75c6c3ec8d73a Binary files /dev/null and b/style_vector_v2/en_US_vctk_p294.wav differ diff --git a/style_vector_v2/en_US_vctk_p295.wav b/style_vector_v2/en_US_vctk_p295.wav new file mode 100644 index 0000000000000000000000000000000000000000..d48a1f4b428ed9da57cb2807d8ef801bfa610b5e Binary files /dev/null and b/style_vector_v2/en_US_vctk_p295.wav differ diff --git a/style_vector_v2/en_US_vctk_p297.wav b/style_vector_v2/en_US_vctk_p297.wav new file mode 100644 index 0000000000000000000000000000000000000000..fbe68c24f78273fc5a7150077255917335bd547a Binary files /dev/null and b/style_vector_v2/en_US_vctk_p297.wav differ diff --git a/style_vector_v2/en_US_vctk_p298.wav b/style_vector_v2/en_US_vctk_p298.wav new file mode 100644 index 0000000000000000000000000000000000000000..a31e261770560da7889fd0573aae33ed43df9c84 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p298.wav differ diff --git a/style_vector_v2/en_US_vctk_p299.wav b/style_vector_v2/en_US_vctk_p299.wav new file mode 100644 index 0000000000000000000000000000000000000000..3946ee528f08704e9d25be63b471813173f52a17 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p299.wav differ diff --git a/style_vector_v2/en_US_vctk_p300.wav b/style_vector_v2/en_US_vctk_p300.wav new file mode 100644 index 0000000000000000000000000000000000000000..b9774283e88b197b735cd7ef3c2e92d190cb7f8d Binary files /dev/null and b/style_vector_v2/en_US_vctk_p300.wav differ diff --git a/style_vector_v2/en_US_vctk_p301.wav b/style_vector_v2/en_US_vctk_p301.wav new file mode 100644 index 0000000000000000000000000000000000000000..01a3e7aa64f488bf6e08c72f8786c1c56c04ef49 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p301.wav differ diff --git a/style_vector_v2/en_US_vctk_p302.wav b/style_vector_v2/en_US_vctk_p302.wav new file mode 100644 index 0000000000000000000000000000000000000000..ab4d8f80f076156078c392975fdbe8943715c5cb Binary files /dev/null and b/style_vector_v2/en_US_vctk_p302.wav differ diff --git a/style_vector_v2/en_US_vctk_p303.wav b/style_vector_v2/en_US_vctk_p303.wav new file mode 100644 index 0000000000000000000000000000000000000000..5a309f8e8fb7f87ccf4a8f084ab356fc5b94a5e8 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p303.wav differ diff --git a/style_vector_v2/en_US_vctk_p304.wav b/style_vector_v2/en_US_vctk_p304.wav new file mode 100644 index 0000000000000000000000000000000000000000..fcd591d8109dc6880656c5013ecc064aca02cf30 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p304.wav differ diff --git a/style_vector_v2/en_US_vctk_p305.wav b/style_vector_v2/en_US_vctk_p305.wav new file mode 100644 index 0000000000000000000000000000000000000000..be068a9fb07fc089156cb691ee2b681eecf34b4f Binary files /dev/null and b/style_vector_v2/en_US_vctk_p305.wav differ diff --git a/style_vector_v2/en_US_vctk_p306.wav b/style_vector_v2/en_US_vctk_p306.wav new file mode 100644 index 0000000000000000000000000000000000000000..58fde7c2a92ded9bc128da1ef33c98872b040350 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p306.wav differ diff --git a/style_vector_v2/en_US_vctk_p307.wav b/style_vector_v2/en_US_vctk_p307.wav new file mode 100644 index 0000000000000000000000000000000000000000..e850009afb0e0894326208aab916d854f0da69df Binary files /dev/null and b/style_vector_v2/en_US_vctk_p307.wav differ diff --git a/style_vector_v2/en_US_vctk_p308.wav b/style_vector_v2/en_US_vctk_p308.wav new file mode 100644 index 0000000000000000000000000000000000000000..0f540a6b3baf8448dd4cc7fbd8a3a5bb623b6117 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p308.wav differ diff --git a/style_vector_v2/en_US_vctk_p310.wav b/style_vector_v2/en_US_vctk_p310.wav new file mode 100644 index 0000000000000000000000000000000000000000..7328e75fd3395bfd6c2ee3326494a7851e3f3f40 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p310.wav differ diff --git a/style_vector_v2/en_US_vctk_p311.wav b/style_vector_v2/en_US_vctk_p311.wav new file mode 100644 index 0000000000000000000000000000000000000000..8545d6eb112b9ac2dcf85688fc2a110fcd3512e2 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p311.wav differ diff --git a/style_vector_v2/en_US_vctk_p312.wav b/style_vector_v2/en_US_vctk_p312.wav new file mode 100644 index 0000000000000000000000000000000000000000..3cc2820f49713ccb344a4a3d9b18d2ac49b2fd9c Binary files /dev/null and b/style_vector_v2/en_US_vctk_p312.wav differ diff --git a/style_vector_v2/en_US_vctk_p313.wav b/style_vector_v2/en_US_vctk_p313.wav new file mode 100644 index 0000000000000000000000000000000000000000..c8e92cbae69a59002bdc3b7c64027c6e6680e044 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p313.wav differ diff --git a/style_vector_v2/en_US_vctk_p314.wav b/style_vector_v2/en_US_vctk_p314.wav new file mode 100644 index 0000000000000000000000000000000000000000..0cea93dd7fa4e660d00d108a52383e9fd517e9e8 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p314.wav differ diff --git a/style_vector_v2/en_US_vctk_p316.wav b/style_vector_v2/en_US_vctk_p316.wav new file mode 100644 index 0000000000000000000000000000000000000000..cfba9cae6cbe3e73231141e8b14bdda5bdbe54f2 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p316.wav differ diff --git a/style_vector_v2/en_US_vctk_p317.wav b/style_vector_v2/en_US_vctk_p317.wav new file mode 100644 index 0000000000000000000000000000000000000000..d97934a87e7266209fdfae11f09947a127d907e5 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p317.wav differ diff --git a/style_vector_v2/en_US_vctk_p318.wav b/style_vector_v2/en_US_vctk_p318.wav new file mode 100644 index 0000000000000000000000000000000000000000..2271784a77f686d98ac3c07fb73db5e5cedca0cd Binary files /dev/null and b/style_vector_v2/en_US_vctk_p318.wav differ diff --git a/style_vector_v2/en_US_vctk_p323.wav b/style_vector_v2/en_US_vctk_p323.wav new file mode 100644 index 0000000000000000000000000000000000000000..df1fa0a98060f8a49dd6078613d66e98b53092bb Binary files /dev/null and b/style_vector_v2/en_US_vctk_p323.wav differ diff --git a/style_vector_v2/en_US_vctk_p326.wav b/style_vector_v2/en_US_vctk_p326.wav new file mode 100644 index 0000000000000000000000000000000000000000..82c4f61a4fd6cd944f68b66379bdfaa48444cc24 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p326.wav differ diff --git a/style_vector_v2/en_US_vctk_p329.wav b/style_vector_v2/en_US_vctk_p329.wav new file mode 100644 index 0000000000000000000000000000000000000000..d811507d533c4431ebcb6380f77961360314142a Binary files /dev/null and b/style_vector_v2/en_US_vctk_p329.wav differ diff --git a/style_vector_v2/en_US_vctk_p330.wav b/style_vector_v2/en_US_vctk_p330.wav new file mode 100644 index 0000000000000000000000000000000000000000..68a9cb9bb3656376e729a78240f3f0ab12f65d42 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p330.wav differ diff --git a/style_vector_v2/en_US_vctk_p333.wav b/style_vector_v2/en_US_vctk_p333.wav new file mode 100644 index 0000000000000000000000000000000000000000..a36dddda2c93bcfc4bb672736a9dd2b4ee182a72 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p333.wav differ diff --git a/style_vector_v2/en_US_vctk_p334.wav b/style_vector_v2/en_US_vctk_p334.wav new file mode 100644 index 0000000000000000000000000000000000000000..03d8f83ee6d3734ca7cbb6179ca9cf16a2e4bff8 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p334.wav differ diff --git a/style_vector_v2/en_US_vctk_p335.wav b/style_vector_v2/en_US_vctk_p335.wav new file mode 100644 index 0000000000000000000000000000000000000000..f4b7926ea26c7430449a12b1330041e5cee8bf79 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p335.wav differ diff --git a/style_vector_v2/en_US_vctk_p336.wav b/style_vector_v2/en_US_vctk_p336.wav new file mode 100644 index 0000000000000000000000000000000000000000..96e011db63174bdab18aa5f668ea70e8e250616a Binary files /dev/null and b/style_vector_v2/en_US_vctk_p336.wav differ diff --git a/style_vector_v2/en_US_vctk_p339.wav b/style_vector_v2/en_US_vctk_p339.wav new file mode 100644 index 0000000000000000000000000000000000000000..1ea96fe856ff39d820d8c6bffe27730d12949c6b Binary files /dev/null and b/style_vector_v2/en_US_vctk_p339.wav differ diff --git a/style_vector_v2/en_US_vctk_p340.wav b/style_vector_v2/en_US_vctk_p340.wav new file mode 100644 index 0000000000000000000000000000000000000000..cda04d08d35f59d131e6e980a3ca888bdce76ee3 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p340.wav differ diff --git a/style_vector_v2/en_US_vctk_p341.wav b/style_vector_v2/en_US_vctk_p341.wav new file mode 100644 index 0000000000000000000000000000000000000000..c515bdca219dca6439af4c1e40b33379fc4dd05a Binary files /dev/null and b/style_vector_v2/en_US_vctk_p341.wav differ diff --git a/style_vector_v2/en_US_vctk_p343.wav b/style_vector_v2/en_US_vctk_p343.wav new file mode 100644 index 0000000000000000000000000000000000000000..acbb2a3519cef2b17dfee856e79f55decbc70071 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p343.wav differ diff --git a/style_vector_v2/en_US_vctk_p345.wav b/style_vector_v2/en_US_vctk_p345.wav new file mode 100644 index 0000000000000000000000000000000000000000..4892f130b521456ea73ac0eb7d59dfef95199b13 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p345.wav differ diff --git a/style_vector_v2/en_US_vctk_p347.wav b/style_vector_v2/en_US_vctk_p347.wav new file mode 100644 index 0000000000000000000000000000000000000000..383faea18c4879c5e71c34c9dc95d133c406acdf Binary files /dev/null and b/style_vector_v2/en_US_vctk_p347.wav differ diff --git a/style_vector_v2/en_US_vctk_p351.wav b/style_vector_v2/en_US_vctk_p351.wav new file mode 100644 index 0000000000000000000000000000000000000000..138f9ebc6a7e23ae8e5aa0e8f0e440533657721d Binary files /dev/null and b/style_vector_v2/en_US_vctk_p351.wav differ diff --git a/style_vector_v2/en_US_vctk_p360.wav b/style_vector_v2/en_US_vctk_p360.wav new file mode 100644 index 0000000000000000000000000000000000000000..2481c090b3e65a26233ce9843516d8dea75df553 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p360.wav differ diff --git a/style_vector_v2/en_US_vctk_p361.wav b/style_vector_v2/en_US_vctk_p361.wav new file mode 100644 index 0000000000000000000000000000000000000000..07ad215840e6a6a0a8d11e6c317960eb833dd563 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p361.wav differ diff --git a/style_vector_v2/en_US_vctk_p362.wav b/style_vector_v2/en_US_vctk_p362.wav new file mode 100644 index 0000000000000000000000000000000000000000..e52bc3749304a9fa446e597f98104bfb4356dff9 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p362.wav differ diff --git a/style_vector_v2/en_US_vctk_p363.wav b/style_vector_v2/en_US_vctk_p363.wav new file mode 100644 index 0000000000000000000000000000000000000000..cfdecdd912fbe1717e6f4dedc04255025ce38b92 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p363.wav differ diff --git a/style_vector_v2/en_US_vctk_p364.wav b/style_vector_v2/en_US_vctk_p364.wav new file mode 100644 index 0000000000000000000000000000000000000000..22baae3a058fa86af14b09c2a0c37db9c2e474a3 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p364.wav differ diff --git a/style_vector_v2/en_US_vctk_p374.wav b/style_vector_v2/en_US_vctk_p374.wav new file mode 100644 index 0000000000000000000000000000000000000000..e61ca0e92ddd8364591a042bffb6012aecc05eca Binary files /dev/null and b/style_vector_v2/en_US_vctk_p374.wav differ diff --git a/style_vector_v2/en_US_vctk_p376.wav b/style_vector_v2/en_US_vctk_p376.wav new file mode 100644 index 0000000000000000000000000000000000000000..c965d551418cb293c699427516dda76a887c99a2 Binary files /dev/null and b/style_vector_v2/en_US_vctk_p376.wav differ diff --git a/style_vector_v2/en_US_vctk_s5.wav b/style_vector_v2/en_US_vctk_s5.wav new file mode 100644 index 0000000000000000000000000000000000000000..627a34c62b6b285a056988ea3a604c8d18b3650c Binary files /dev/null and b/style_vector_v2/en_US_vctk_s5.wav differ diff --git a/tts_harvard.py b/tts_harvard.py index 8f404cd453eaaf0c450ffe4e1559652ca646c558..2b06aa004b6716ba764bdb9fe0bf194b3e1c1cda 100644 --- a/tts_harvard.py +++ b/tts_harvard.py @@ -1,13 +1,7 @@ # Synthesize all Harvard Lists 77x lists of 10x sentences to single .wav -# -# 1. using mimic3 style -# Folder: 'prompt_mimic3/' -# 2. using mimic3 4x accelerated style -# Folder: 'prompt_mimic3speed/' -# 3. using crema-d style -# Folder: 'prompt_human/' -# -# WAVS used from tts_paper_plot.py + +# 1. using mimic3 english 1x/4x non-english 1x/4x +# Use visualize_tts_plesantness.py for figs --> 4figs eng 1x 4x vs human non-eng 1x 4x vs human libri import soundfile import json @@ -69,27 +63,10 @@ def load_speech(split=None): -# Generate 77 wavs - - - - -with open('voices.json', 'r') as f: - df = json.load(f)['voices'] -voice_names = [v['voice'] for k,v in df.items()] -synthetic_wav_paths = [] -synthetic_wav_paths_AFFECT = [] -for voice in voice_names: - synthetic_wav_paths.append( - '/data/dkounadis/shift/assets/wavs/style_vector/' + voice.replace('/', '_').replace('#', '_').replace( - 'cmu-arctic', 'cmu_arctic').replace('_low', '') + '.wav') - synthetic_wav_paths_AFFECT.append( - '/data/dkounadis/shift/assets/wavs/style_vector_v2/' + voice.replace('/', '_').replace('#', '_').replace( - 'cmu-arctic', 'cmu_arctic').replace('_low', '') + '.wav') -print(len(synthetic_wav_paths)) + natural_wav_paths = load_speech() @@ -97,47 +74,66 @@ natural_wav_paths = load_speech() # SYNTHESIZE mimic mimicx4 crema-d import msinference - +import os with open('harvard.json', 'r') as f: harvard_individual_sentences = json.load(f)['sentences'] - - - -for audio_prompt in ['mimic3', 'mimic3_speed', 'human']: - total_audio = [] - ix = 0 - for list_of_10 in harvard_individual_sentences: - # long_sentence = ' '.join(list_of_10['sentences']) - # harvard.append(long_sentence.replace('.', ' ')) - for text in list_of_10['sentences']: - if audio_prompt == 'mimic3': - style_vec = msinference.compute_style( - synthetic_wav_paths[ix % 134]) - elif audio_prompt == 'mimic3_speed': - style_vec = msinference.compute_style( - synthetic_wav_paths_AFFECT[ix % 134]) - elif audio_prompt == 'human': - style_vec = msinference.compute_style( - natural_wav_paths[ix % len(natural_wav_paths)]) - else: - print('unknonw list of style vecto') - print(ix, text) - ix += 1 - x = msinference.inference(text, - style_vec, - alpha=0.3, - beta=0.7, - diffusion_steps=7, - embedding_scale=1) - - total_audio.append(x) - # concat before write - # -- for 10x sentenctes - print('_____________________') - # -- for 77x lists - total_audio = np.concatenate(total_audio) - soundfile.write(f'{audio_prompt}_770.wav', total_audio, 24000) - print(f'{audio_prompt}_full_770.wav') \ No newline at end of file +synthetic_wav_paths = ['./style_vector/' + i for i in + os.listdir('./style_vector/')] +synthetic_wav_paths_4x = ['./style_vector_v2/' + i for i in + os.listdir('./style_vector_v2/')] +synthetic_wav_paths_foreign = ['./mimic3_foreign/' + i for i in os.listdir('./mimic3_foreign/')] +synthetic_wav_paths_foreign_4x = ['./mimic3_foreign_4x/' + i for i in os.listdir('./mimic3_foreign_4x/')] + + + +for audio_prompt in ['mimic3', + 'mimic3_speed', + 'human', + 'foreign', + 'foreign_speed']: + if not os.path.isfile(f'{audio_prompt}_k.wav'): + total_audio = [] + ix = 0 + for list_of_10 in harvard_individual_sentences[:2]: + # long_sentence = ' '.join(list_of_10['sentences']) + # harvard.append(long_sentence.replace('.', ' ')) + for text in list_of_10['sentences']: + if audio_prompt == 'mimic3': + style_vec = msinference.compute_style( + synthetic_wav_paths[ix % 134]) + elif audio_prompt == 'mimic3_speed': + style_vec = msinference.compute_style( + synthetic_wav_paths_4x[ix % 134]) + elif audio_prompt == 'human': + style_vec = msinference.compute_style( + natural_wav_paths[ix % len(natural_wav_paths)]) + elif audio_prompt == 'foreign': + style_vec = msinference.compute_style( + synthetic_wav_paths_foreign[ix % 204]) + elif audio_prompt == 'foreign_speed': + style_vec = msinference.compute_style( + synthetic_wav_paths_foreign_4x[ix % 204]) + else: + print('unknonw list of style vecto') + print(ix, text) + ix += 1 + x = msinference.inference(text, + style_vec, + alpha=0.3, + beta=0.7, + diffusion_steps=7, + embedding_scale=1) + + total_audio.append(x) + # concat before write + # -- for 10x sentenctes + print('_____________________') + # -- for 77x lists + total_audio = np.concatenate(total_audio) + soundfile.write(f'{audio_prompt}_k.wav', total_audio, 24000) + + else: + print('\nALREADY EXISTS\n') \ No newline at end of file diff --git a/visualize_tts_plesantness.py b/visualize_tts_plesantness.py index 254d459be64a8e418ab98ac071dab3e2b817f73f..d0fbaee662523d6a2631342af9c60b0e0e9eba78 100644 --- a/visualize_tts_plesantness.py +++ b/visualize_tts_plesantness.py @@ -25,7 +25,7 @@ import matplotlib.pyplot as plt import audiofile LABELS = ['arousal', 'dominance', 'valence', - 'speech_synthesizer', 'synthetic_singing', + # 'speech_synthesizer', 'synthetic_singing', 'Angry', 'Sad', 'Happy', @@ -37,9 +37,9 @@ LABELS = ['arousal', 'dominance', 'valence', ] -args = transformers.Wav2Vec2Config() #finetuning_task='spef2feat_reg') -args.dev = torch.device('cuda:0') -args.dev2 = torch.device('cuda:0') +config = transformers.Wav2Vec2Config() #finetuning_task='spef2feat_reg') +config.dev = torch.device('cuda:0') +config.dev2 = torch.device('cuda:0') # def _softmax(x): # '''x : (batch, num_class)''' @@ -81,11 +81,11 @@ def _sigmoid(x): # for mimic3/mimic3speed/human - concat all 77 and run timeseries with 7s hop 3s for long_audio in [ - # 'mimic3.wav', - # 'mimic3_speedup.wav', - 'human_770.wav', # 'mimic3_all_77.wav', # - 'mimic3_770.wav', - 'mimic3_speed_770.wav' + 'mimic3_k.wav', + 'mimic_speed_k.wav', + 'human_k.wav' + 'foreign_k.wav', + 'foreign_speed_k.wav', ]: file_interface = f'timeseries_{long_audio.replace("/", "")}.pkl' if not os.path.exists(file_interface): @@ -118,50 +118,120 @@ for long_audio in [ teacher_cat = AutoModelForAudioClassification.from_pretrained( '3loi/SER-Odyssey-Baseline-WavLM-Categorical-Attributes', trust_remote_code=True # fun definitions see 3loi/SER-.. repo - ).to(args.dev2).eval() + ).to(config.dev2).eval() teacher_cat.forward = types.MethodType(_infer, teacher_cat) - - # Audioset & ADV + # ===================[:]===================== Dawn + def _prenorm(x, attention_mask=None): + '''mean/var''' + if attention_mask is not None: + N = attention_mask.sum(1, keepdim=True) # here attn msk is unprocessed just the original input + x -= x.sum(1, keepdim=True) / N + var = (x * x).sum(1, keepdim=True) / N + + else: + x -= x.mean(1, keepdim=True) # mean is an onnx operator reducemean saves some ops compared to casting integer N to float and the div + var = (x * x).mean(1, keepdim=True) + return x / torch.sqrt(var + 1e-7) + + from torch import nn + from transformers.models.wav2vec2.modeling_wav2vec2 import Wav2Vec2PreTrainedModel, Wav2Vec2Model + class RegressionHead(nn.Module): + r"""Classification head.""" + + def __init__(self, config): + + super().__init__() + + self.dense = nn.Linear(config.hidden_size, config.hidden_size) + self.dropout = nn.Dropout(config.final_dropout) + self.out_proj = nn.Linear(config.hidden_size, config.num_labels) + + def forward(self, features, **kwargs): + + x = features + x = self.dropout(x) + x = self.dense(x) + x = torch.tanh(x) + x = self.dropout(x) + x = self.out_proj(x) + + return x + + + class Dawn(Wav2Vec2PreTrainedModel): + r"""Speech emotion classifier.""" + + def __init__(self, config): + + super().__init__(config) + + self.config = config + self.wav2vec2 = Wav2Vec2Model(config) + self.classifier = RegressionHead(config) + self.init_weights() + + def forward( + self, + input_values, + attention_mask=None, + ): + x = _prenorm(input_values, attention_mask=attention_mask) + outputs = self.wav2vec2(x, attention_mask=attention_mask) + hidden_states = outputs[0] + hidden_states = torch.mean(hidden_states, dim=1) + logits = self.classifier(hidden_states) + return logits + # return {'hidden_states': hidden_states, + # 'logits': logits} + dawn = Dawn.from_pretrained('audeering/wav2vec2-large-robust-12-ft-emotion-msp-dim').to(config.dev).eval() + # ======================================= + + + + + + - audioset_model = audonnx.load(audmodel.load('17c240ec-1.0.0'), device='cuda:0') - adv_model = audonnx.load(audmodel.load('90398682-2.0.0'), device='cuda:0') - def process_function(x, sampling_rate, idx): '''run audioset ct, adv - USE onnx teachers - - return [synth-speech, synth-singing, 7x, 3x adv] = 11 + USE onnx teachers + + return [synth-speech, synth-singing, 7x, 3x adv] = 11 ''' # x = x[None , :] ASaHSuFDCN #{0: 'Angry', 1: 'Sad', 2: 'Happy', 3: 'Surprise', #4: 'Fear', 5: 'Disgust', 6: 'Contempt', 7: 'Neutral'} #tensor([[0.0015, 0.3651, 0.0593, 0.0315, 0.0600, 0.0125, 0.0319, 0.4382]]) - logits_cat = teacher_cat(torch.from_numpy(x).to(args.dev)).cpu().detach().numpy() + logits_cat = teacher_cat(torch.from_numpy(x).to(config.dev)).cpu().detach().numpy() # USE ALL CATEGORIES # -- - logits_audioset = audioset_model(x, 16000)['logits_sounds'] - logits_audioset = logits_audioset[:, [7, 35]] # speech synthesizer synthetic singing + # logits_audioset = audioset_model(x, 16000)['logits_sounds'] + # logits_audioset = logits_audioset[:, [7, 35]] # speech synthesizer synthetic singing # -- - logits_adv = adv_model(x, 16000)['logits'] + logits_adv = dawn(torch.from_numpy(x).to(config.dev)).cpu().detach().numpy() #['logits'] cat = np.concatenate([logits_adv, - _sigmoid(logits_audioset), - _softmax(logits_cat)], - 1) + # _sigmoid(logits_audioset), + _softmax(logits_cat)], + 1) print(cat) - return cat #logits_adv #model(signal, sampling_rate)['logits'] + return cat #logits_adv #model(signal, sampling_rate)['logits'] + + +# --------------------- + interface = audinterface.Feature( feature_names=LABELS, process_func=process_function, # process_func_args={'outputs': 'logits_scene'}, process_func_applies_sliding_window=False, - win_dur=40.0, - hop_dur=10.0, + win_dur=4.0, + hop_dur=40.0, sampling_rate=16000, resample=True, verbose=True, @@ -208,7 +278,7 @@ for k,v in preds.items(): # Show plots by 2 -fig, ax = plt.subplots(nrows=10, ncols=2, figsize=(24, 24), gridspec_kw={'hspace': 0, 'wspace': .04}) +fig, ax = plt.subplots(nrows=8, ncols=2, figsize=(4.6, 24), gridspec_kw={'hspace': 0, 'wspace': .04}) # ADV @@ -294,10 +364,10 @@ time_stamp = preds['human_770.wav'].index.to_numpy() for j, dim in enumerate(['Angry', 'Sad', 'Happy', - 'Surprise', + # 'Surprise', 'Fear', 'Disgust', - 'Contempt', + # 'Contempt', # 'Neutral' ]): # ASaHSuFDCN j = j + 3 # skip A/D/V suplt @@ -366,6 +436,6 @@ for j, dim in enumerate(['Angry', -plt.savefig(f'valence_tts.pdf', bbox_inches='tight') +plt.savefig(f'fig8.pdf', bbox_inches='tight') plt.close()