jonatasgrosman/exp_w2v2r_de_xls-r_age_teens-10_sixties-0_s460

.gitattributes

@@ -32,3 +32,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+cv7_transcriptions.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,17 @@
+---
+language:
+- de
+license: apache-2.0
+tags:
+- automatic-speech-recognition
+- de
+datasets:
+- mozilla-foundation/common_voice_7_0
+---
+# exp_w2v2r_de_xls-r_age_teens-10_sixties-0_s460
+
+Fine-tuned [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) for speech recognition using the train split of [Common Voice 7.0 (de)](https://huggingface.co/datasets/mozilla-foundation/common_voice_7_0).
+When using this model, make sure that your speech input is sampled at 16kHz.
+
+This model has been fine-tuned by the [HuggingSound](https://github.com/jonatasgrosman/huggingsound) tool.
+    

config.json

@@ -0,0 +1,108 @@
+{
+  "_name_or_path": "facebook/wav2vec2-xls-r-300m",
+  "activation_dropout": 0.05,
+  "adapter_kernel_size": 3,
+  "adapter_stride": 2,
+  "add_adapter": false,
+  "apply_spec_augment": true,
+  "architectures": [
+    "Wav2Vec2ForCTC"
+  ],
+  "attention_dropout": 0.05,
+  "bos_token_id": 1,
+  "classifier_proj_size": 256,
+  "codevector_dim": 768,
+  "contrastive_logits_temperature": 0.1,
+  "conv_bias": true,
+  "conv_dim": [
+    512,
+    512,
+    512,
+    512,
+    512,
+    512,
+    512
+  ],
+  "conv_kernel": [
+    10,
+    3,
+    3,
+    3,
+    3,
+    2,
+    2
+  ],
+  "conv_stride": [
+    5,
+    2,
+    2,
+    2,
+    2,
+    2,
+    2
+  ],
+  "ctc_loss_reduction": "mean",
+  "ctc_zero_infinity": false,
+  "diversity_loss_weight": 0.1,
+  "do_stable_layer_norm": true,
+  "eos_token_id": 2,
+  "feat_extract_activation": "gelu",
+  "feat_extract_dropout": 0.0,
+  "feat_extract_norm": "layer",
+  "feat_proj_dropout": 0.05,
+  "feat_quantizer_dropout": 0.0,
+  "final_dropout": 0.05,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0.05,
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "layer_norm_eps": 1e-05,
+  "layerdrop": 0.05,
+  "mask_feature_length": 10,
+  "mask_feature_min_masks": 0,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_masks": 2,
+  "mask_time_prob": 0.05,
+  "model_type": "wav2vec2",
+  "num_adapter_layers": 3,
+  "num_attention_heads": 16,
+  "num_codevector_groups": 2,
+  "num_codevectors_per_group": 320,
+  "num_conv_pos_embedding_groups": 16,
+  "num_conv_pos_embeddings": 128,
+  "num_feat_extract_layers": 7,
+  "num_hidden_layers": 24,
+  "num_negatives": 100,
+  "output_hidden_size": 1024,
+  "pad_token_id": 0,
+  "proj_codevector_dim": 768,
+  "tdnn_dilation": [
+    1,
+    2,
+    3,
+    1,
+    1
+  ],
+  "tdnn_dim": [
+    512,
+    512,
+    512,
+    512,
+    1500
+  ],
+  "tdnn_kernel": [
+    5,
+    3,
+    3,
+    1,
+    1
+  ],
+  "torch_dtype": "float32",
+  "transformers_version": "4.18.0",
+  "use_weighted_layer_sum": false,
+  "vocab_size": 38,
+  "xvector_output_dim": 512
+}

cv7_evaluation.json

@@ -0,0 +1,396 @@
+{
+  "validation": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.04547381716627499,
+      "wer": 0.18391866913123844
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.04475424069188274,
+      "wer": 0.18243125109273922
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.04836098745447187,
+      "wer": 0.1926260346124906
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.04150526843953838,
+      "wer": 0.16419956140350878
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.04706651413629678,
+      "wer": 0.18496464458503908
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.05254112447827154,
+      "wer": 0.2108843537414966
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.06382623224728488,
+      "wer": 0.22574257425742575
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.053833605220228384,
+      "wer": 0.2073170731707317
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.046275783980702014,
+      "wer": 0.19184086311530682
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.054394545724025495,
+      "wer": 0.2158912702561422
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.044751405132793166,
+      "wer": 0.1824161185579885
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.048927361686112156,
+      "wer": 0.1899948159668222
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.07936507936507936,
+      "wer": 0.29878048780487804
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.048478338248754194,
+      "wer": 0.18992180712423978
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.04545591850581603,
+      "wer": 0.18415871816918045
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.09926470588235294,
+      "wer": 0.425
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.037162162162162164,
+      "wer": 0.20833333333333334
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.032432432432432434,
+      "wer": 0.15384615384615385
+    },
+    {
+      "subset_type": "accent",
+      "subset": "slovakia",
+      "cer": 0.030555555555555555,
+      "wer": 0.15789473684210525
+    },
+    {
+      "subset_type": "accent",
+      "subset": "netherlands",
+      "cer": 0.04040404040404041,
+      "wer": 0.1956521739130435
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.060019965061143,
+      "wer": 0.22711571675302245
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.07869249394673124,
+      "wer": 0.256198347107438
+    },
+    {
+      "subset_type": "accent",
+      "subset": "czechia",
+      "cer": 0.04452054794520548,
+      "wer": 0.15757575757575756
+    },
+    {
+      "subset_type": "accent",
+      "subset": "turkey",
+      "cer": 0.11793611793611794,
+      "wer": 0.4067796610169492
+    },
+    {
+      "subset_type": "accent",
+      "subset": "russia",
+      "cer": 0.07337883959044368,
+      "wer": 0.24598930481283424
+    },
+    {
+      "subset_type": "accent",
+      "subset": "austria",
+      "cer": 0.04709610849056604,
+      "wer": 0.19847715736040608
+    },
+    {
+      "subset_type": "accent",
+      "subset": "poland",
+      "cer": 0.02926829268292683,
+      "wer": 0.12121212121212122
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.09067357512953368,
+      "wer": 0.29914529914529914
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.07065868263473053,
+      "wer": 0.224
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.04269625281158773,
+      "wer": 0.17276912957684717
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.08933263268523384,
+      "wer": 0.3286219081272085
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.052414106356944976,
+      "wer": 0.19950410725569384
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.0516246759794233,
+      "wer": 0.1974016019622121
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.05474440945382829,
+      "wer": 0.21170879239576582
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.045536034420939404,
+      "wer": 0.18957816377171216
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.06396416244400382,
+      "wer": 0.22945891783567135
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.06928519151012442,
+      "wer": 0.2470389170896785
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.08023132687138551,
+      "wer": 0.2626038781163435
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.07522935779816514,
+      "wer": 0.2876712328767123
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.052448988907655085,
+      "wer": 0.19282238442822383
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.08142246404974737,
+      "wer": 0.26975476839237056
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.051714102852861224,
+      "wer": 0.19761864627549536
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.07083610825210389,
+      "wer": 0.24916247906197656
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.1254180602006689,
+      "wer": 0.26136363636363635
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.05519457588812769,
+      "wer": 0.2088193002077009
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.05229948747895832,
+      "wer": 0.19936780997826847
+    },
+    {
+      "subset_type": "accent",
+      "subset": "austria",
+      "cer": 0.05788906838987615,
+      "wer": 0.2123730378578024
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.06771463119709795,
+      "wer": 0.22406639004149378
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.08121827411167512,
+      "wer": 0.3333333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.0528169014084507,
+      "wer": 0.21052631578947367
+    },
+    {
+      "subset_type": "accent",
+      "subset": "denmark",
+      "cer": 0.04411764705882353,
+      "wer": 0.3333333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "czechia",
+      "cer": 0.17615176151761516,
+      "wer": 0.5471698113207547
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.11598746081504702,
+      "wer": 0.37254901960784315
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.07274490785645005,
+      "wer": 0.2733333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "slovakia",
+      "cer": 0.050505050505050504,
+      "wer": 0.1724137931034483
+    },
+    {
+      "subset_type": "accent",
+      "subset": "bulgaria",
+      "cer": 0.0,
+      "wer": 0.0
+    },
+    {
+      "subset_type": "accent",
+      "subset": "russia",
+      "cer": 0.21548821548821548,
+      "wer": 0.4666666666666667
+    },
+    {
+      "subset_type": "accent",
+      "subset": "greece",
+      "cer": 0.1282051282051282,
+      "wer": 0.25925925925925924
+    },
+    {
+      "subset_type": "accent",
+      "subset": "netherlands",
+      "cer": 0.06354009077155824,
+      "wer": 0.26424870466321243
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.07144306880454637,
+      "wer": 0.2647058823529412
+    },
+    {
+      "subset_type": "accent",
+      "subset": "turkey",
+      "cer": 0.15053763440860216,
+      "wer": 0.4
+    },
+    {
+      "subset_type": "accent",
+      "subset": "poland",
+      "cer": 0.10367892976588629,
+      "wer": 0.34782608695652173
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.010869565217391304,
+      "wer": 0.07142857142857142
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.04868457637666,
+      "wer": 0.18658734451054623
+    },
+    {
+      "subset_type": "accent",
+      "subset": "hungary",
+      "cer": 0.09230769230769231,
+      "wer": 0.42857142857142855
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e048b92bb0ec2e91f722f83ed3f8c942a4d3edacdc1dcebcb04803dd90db2c6
+size 139019999

preprocessor_config.json

@@ -0,0 +1,10 @@
+{
+  "do_normalize": true,
+  "feature_extractor_type": "Wav2Vec2FeatureExtractor",
+  "feature_size": 1,
+  "padding_side": "right",
+  "padding_value": 0,
+  "processor_class": "Wav2Vec2Processor",
+  "return_attention_mask": true,
+  "sampling_rate": 16000
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be87c2c7907c1f65b4de41f6dd71d2a383c95a9b51dd428d9e909c917fec7844
+size 1262054513

special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>"}

tokenizer_config.json

@@ -0,0 +1 @@
+{"unk_token": "<unk>", "bos_token": "<s>", "eos_token": "</s>", "pad_token": "<pad>", "do_lower_case": false, "word_delimiter_token": "|", "replace_word_delimiter_char": " ", "special_tokens_map_file": null, "name_or_path": "/tmp/tmp439sqd1n", "tokenizer_class": "Wav2Vec2CTCTokenizer", "processor_class": "Wav2Vec2Processor"}

vocab.json

@@ -0,0 +1 @@
+{"<pad>": 0, "<s>": 1, "</s>": 2, "<unk>": 3, "|": 4, "'": 5, "-": 6, "a": 7, "b": 8, "c": 9, "d": 10, "e": 11, "f": 12, "g": 13, "h": 14, "i": 15, "j": 16, "k": 17, "l": 18, "m": 19, "n": 20, "o": 21, "p": 22, "q": 23, "r": 24, "s": 25, "t": 26, "u": 27, "v": 28, "w": 29, "x": 30, "y": 31, "z": 32, "ß": 33, "á": 34, "ä": 35, "ö": 36, "ü": 37}