File size: 3,984 Bytes
b5b9f4f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b29b819
b5b9f4f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b29b819
 
 
b5b9f4f
 
 
 
 
b29b819
b5b9f4f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
0b6a34f
b5b9f4f
 
 
 
 
 
 
 
b29b819
b5b9f4f
 
 
 
 
f373715
 
 
 
b5b9f4f
 
 
 
 
 
f373715
b5b9f4f
f373715
 
b5b9f4f
f373715
 
 
b5b9f4f
 
 
 
 
 
b29b819
b5b9f4f
 
 
 
 
 
 
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
---
language: ka 
datasets:
- common_voice
metrics:
- wer
tags:
- audio
- automatic-speech-recognition
- speech
- xlsr-fine-tuning-week
license: apache-2.0
model-index:
- name: XLSR Wav2Vec finetuned for Georgian
  results:
  - task: 
      name: Speech Recognition
      type: automatic-speech-recognition
    dataset:
      name: Common Voice ka 
      type: common_voice
      args: ka
    metrics:
       - name: Test WER
         type: wer
         value: 45.28
---

# Wav2Vec2-Large-XLSR-53-Georgian

Fine-tuned [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on Georgian using the [Common Voice](https://huggingface.co/datasets/common_voice).
When using this model, make sure that your speech input is sampled at 16kHz.

## Usage

The model can be used directly (without a language model) as follows:

```python
import librosa
import torch
import torchaudio
from datasets import load_dataset
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor


test_dataset = load_dataset("common_voice", "ka", split="test[:2%]") 

processor = Wav2Vec2Processor.from_pretrained("xsway/wav2vec2-large-xlsr-georgian")
model = Wav2Vec2ForCTC.from_pretrained("xsway/wav2vec2-large-xlsr-georgian") 

resampler = lambda sampling_rate, y: librosa.resample(y.numpy().squeeze(), sampling_rate, 16_000)

# Preprocessing the datasets.
# We need to read the audio files as arrays
def speech_file_to_array_fn(batch):
\\\\tspeech_array, sampling_rate = torchaudio.load(batch["path"])
\\\\tbatch["speech"] = resampler(sampling_rate, speech_array).squeeze()
\\\\treturn batch

test_dataset = test_dataset.map(speech_file_to_array_fn)
inputs = processor(test_dataset["speech"][:2], sampling_rate=16_000, return_tensors="pt", padding=True)

with torch.no_grad():
\\\\tlogits = model(inputs.input_values, attention_mask=inputs.attention_mask).logits

predicted_ids = torch.argmax(logits, dim=-1)

print("Prediction:", processor.batch_decode(predicted_ids))
print("Reference:", test_dataset["sentence"][:2])
```


## Evaluation

The model can be evaluated as follows on the Georgian test data of Common Voice.  


```python
import torch
import torchaudio
from datasets import load_dataset, load_metric
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
import re
import librosa

test_dataset = load_dataset("common_voice", "ka", split="test") 
wer = load_metric("wer")

processor = Wav2Vec2Processor.from_pretrained("xsway/wav2vec2-large-xlsr-georgian") 
model = Wav2Vec2ForCTC.from_pretrained("xsway/wav2vec2-large-xlsr-georgian") 
model.to("cuda")

chars_to_ignore_regex = '[\\\\\\\\,\\\\\\\\?\\\\\\\\.\\\\\\\\!\\\\\\\\-\\\\\\\\;\\\\\\\\:\\\\\\\\"\\\\\\\\“]' 
resampler = lambda sampling_rate, y: librosa.resample(y.numpy().squeeze(), sampling_rate, 16_000)

# Preprocessing the datasets.
# We need to read the audio files as arrays
def speech_file_to_array_fn(batch):
  batch["sentence"] = re.sub(chars_to_ignore_regex, '', batch["sentence"]).lower()
  speech_array, sampling_rate = torchaudio.load(batch["path"])
  batch["speech"] = resampler(sampling_rate, speech_array).squeeze()
  return batch

test_dataset = test_dataset.map(speech_file_to_array_fn)

# Preprocessing the datasets.
# We need to read the audio files as arrays
def evaluate(batch):
  inputs = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)

  with torch.no_grad():
    logits = model(inputs.input_values.to("cuda"), attention_mask=inputs.attention_mask.to("cuda")).logits

  pred_ids = torch.argmax(logits, dim=-1)
  batch["pred_strings"] = processor.batch_decode(pred_ids)
  return batch

result = test_dataset.map(evaluate, batched=True, batch_size=8)

print("WER: {:2f}".format(100 * wer.compute(predictions=result["pred_strings"], references=result["sentence"])))
```

**Test Result**: 45.28 %  


## Training

The Common Voice `train`, `validation` datasets were used for training.

The script used for training can be found [here](...)