File size: 5,947 Bytes
a8f9c20
 
 
 
 
 
 
6f59045
a8f9c20
 
 
 
 
 
 
 
6f59045
a8f9c20
0e3b9d4
 
a8f9c20
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
0e3b9d4
a8f9c20
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
0e3b9d4
 
 
 
 
 
 
 
 
 
a8f9c20
6f59045
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
a8f9c20
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
---
license: apache-2.0
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: whisper-small-spanish
  results: []
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# whisper-small-sp

This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the `commonvoice dataset v11` dataset.
It achieves the following results on the evaluation set:
- Loss: 0.4485
- Wer: 20.6842

## Model description

More information needed

## Intended uses & limitations

More information needed

## Training and evaluation data

More information needed

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 0.0005
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 25000
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step  | Validation Loss | Wer     |
|:-------------:|:-----:|:-----:|:---------------:|:-------:|
| 2.2671        | 0.13  | 1000  | 2.2108          | 76.2667 |
| 1.4465        | 0.26  | 2000  | 1.6057          | 67.8753 |
| 1.0997        | 0.39  | 3000  | 1.1928          | 54.2433 |
| 0.9389        | 0.52  | 4000  | 1.0020          | 47.8307 |
| 0.7881        | 0.65  | 5000  | 0.8933          | 46.0046 |
| 0.7596        | 0.78  | 6000  | 0.7721          | 38.5595 |
| 0.5678        | 0.91  | 7000  | 0.6903          | 36.2897 |
| 0.4412        | 1.04  | 8000  | 0.6476          | 32.7473 |
| 0.4239        | 1.17  | 9000  | 0.5973          | 30.8142 |
| 0.3935        | 1.3   | 10000 | 0.5444          | 29.0208 |
| 0.3307        | 1.43  | 11000 | 0.5024          | 27.0434 |
| 0.2937        | 1.56  | 12000 | 0.4608          | 24.7318 |
| 0.2471        | 1.69  | 13000 | 0.4259          | 22.8940 |
| 0.2357        | 1.82  | 14000 | 0.3936          | 21.6018 |
| 0.2292        | 1.95  | 15000 | 0.3776          | 20.8004 |
| 0.1493        | 2.08  | 16000 | 0.4599          | 24.0491 |
| 0.1708        | 2.21  | 17000 | 0.4370          | 23.3443 |
| 0.1385        | 2.34  | 18000 | 0.4277          | 22.3171 |
| 0.1288        | 2.47  | 19000 | 0.4050          | 21.0118 |
| 0.1627        | 2.6   | 20000 | 0.4507          | 23.4004 |
| 0.1675        | 2.73  | 21000 | 0.4346          | 22.8261 |
| 0.159         | 2.86  | 22000 | 0.4179          | 22.2949 |
| 0.1458        | 2.99  | 23000 | 0.3978          | 21.0810 |
| 0.0487        | 3.12  | 24000 | 0.4456          | 20.8617 |
| 0.0401        | 3.25  | 25000 | 0.4485          | 20.6842 |

### Transcription:

```python
from datasets import load_dataset, Audio
import torch
from transformers import WhisperProcessor, WhisperForConditionalGeneration

# device
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

# load the model
processor = WhisperProcessor.from_pretrained("clu-ling/whisper-small-spanish")
model = WhisperForConditionalGeneration.from_pretrained("clu-ling/whisper-small-spanish").to(device)
forced_decoder_ids = processor.get_decoder_prompt_ids(language="es", task="transcribe")

# load the dataset
commonvoice_eval = load_dataset("mozilla-foundation/common_voice_11_0", "es", split="validation", streaming=True)
commonvoice_eval = commonvoice_eval.cast_column("audio", Audio(sampling_rate=16000))
sample = next(iter(commonvoice_eval))["audio"]

# features and generate token ids
input_features = processor(sample["array"], sampling_rate=sample["sampling_rate"], return_tensors="pt").input_features
predicted_ids = model.generate(input_features.to(device), forced_decoder_ids=forced_decoder_ids)

# decode
transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)

print(transcription)

```

### Evaluation:

Evaluates this model on `mozilla-foundation/common_voice_11_0` test split.

```python
from transformers.models.whisper.english_normalizer import BasicTextNormalizer
from datasets import load_dataset, Audio
import evaluate
import torch
import re
from transformers import WhisperProcessor, WhisperForConditionalGeneration

# device
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

# metric
wer_metric = evaluate.load("wer")

# model
processor = WhisperProcessor.from_pretrained("clu-ling/whisper-small-spanish")
model = WhisperForConditionalGeneration.from_pretrained("clu-ling/whisper-small-spanish")

# dataset
dataset = load_dataset("mozilla-foundation/common_voice_11_0", "es", split="test", )#cache_dir=args.cache_dir
dataset = dataset.cast_column("audio", Audio(sampling_rate=16000))

#for debuggings: it gets some examples
#dataset = dataset.shard(num_shards=10000, index=0)
#print(dataset)
   
def normalize(batch):
  batch["gold_text"] = whisper_norm(batch['sentence'])
  return batch

def map_wer(batch):
  model.to(device)
  forced_decoder_ids = processor.get_decoder_prompt_ids(language = "es", task = "transcribe")
  inputs = processor(batch["audio"]["array"], sampling_rate=batch["audio"]["sampling_rate"], return_tensors="pt").input_features
  with torch.no_grad():
    generated_ids = model.generate(inputs=inputs.to(device), forced_decoder_ids=forced_decoder_ids)
    transcription = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
  batch["predicted_text"] = whisper_norm(transcription)
  return batch

# process GOLD text
processed_dataset = dataset.map(normalize)
# get predictions
predicted = processed_dataset.map(map_wer)

# word error rate
wer = wer_metric.compute(references=predicted['gold_text'], predictions=predicted['predicted_text'])
wer = round(100 * wer, 2)
print("WER:", wer)


```

### Framework versions

- Transformers 4.25.1
- Pytorch 1.13.0+cu117
- Datasets 2.7.1
- Tokenizers 0.13.2