1
---
2
language: kn
3
datasets:
4
- openslr
5
metrics:
6
- wer
7
tags:
8
- audio
9
- automatic-speech-recognition
10
- speech
11
- xlsr-fine-tuning-week
12
license: apache-2.0
13
model-index:
14
- name: XLSR Wav2Vec2 Large 53 Kannada by Amogh Gopadi
15
  results:
16
  - task: 
17
      name: Speech Recognition
18
      type: automatic-speech-recognition
19
    dataset:
20
      name: OpenSLR kn
21
      type: openslr
22
    metrics:
23
       - name: Test WER
24
         type: wer
25
         value: 27.08
26
27
---
28
29
# Wav2Vec2-Large-XLSR-53-Kannada
30
31
Fine-tuned [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on Kannada using the [OpenSLR SLR79](http://openslr.org/79/) dataset. When using this model, make sure that your speech input is sampled at 16kHz. 
32
33
## Usage
34
35
The model can be used directly (without a language model) as follows, assuming you have a dataset with Kannada `sentence` and `path` fields:
36
37
```python
38
39
import torch
40
import torchaudio
41
from datasets import load_dataset
42
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
43
44
# test_dataset = #TODO: WRITE YOUR CODE TO LOAD THE TEST DATASET. For a sample, see the Colab link in Training Section.
45
46
processor = Wav2Vec2Processor.from_pretrained("amoghsgopadi/wav2vec2-large-xlsr-kn")
47
model = Wav2Vec2ForCTC.from_pretrained("amoghsgopadi/wav2vec2-large-xlsr-kn")
48
resampler = torchaudio.transforms.Resample(48_000, 16_000) # The original data was with 48,000 sampling rate. You can change it according to your input.
49
50
# Preprocessing the datasets.
51
# We need to read the audio files as arrays
52
def speech_file_to_array_fn(batch):
53
    speech_array, sampling_rate = torchaudio.load(batch["path"])
54
    batch["speech"] = resampler(speech_array).squeeze().numpy()
55
    return batch
56
57
test_dataset = test_dataset.map(speech_file_to_array_fn)
58
inputs = processor(test_dataset["speech"][:2], sampling_rate=16_000, return_tensors="pt", padding=True)
59
60
with torch.no_grad():
61
    logits = model(inputs.input_values, attention_mask=inputs.attention_mask).logits
62
63
predicted_ids = torch.argmax(logits, dim=-1)
64
65
print("Prediction:", processor.batch_decode(predicted_ids))
66
print("Reference:", test_dataset["sentence"][:2])
67
68
```
69
70
## Evaluation
71
72
The model can be evaluated as follows on 10% of the Kannada data on OpenSLR.
73
74
```python
75
76
import torch
77
import torchaudio
78
from datasets import load_dataset, load_metric
79
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
80
import re
81
82
# test_dataset = #TODO: WRITE YOUR CODE TO LOAD THE TEST DATASET. For sample see the Colab link in Training Section.
83
84
wer = load_metric("wer")
85
86
processor = Wav2Vec2Processor.from_pretrained("amoghsgopadi/wav2vec2-large-xlsr-kn")
87
model = Wav2Vec2ForCTC.from_pretrained("amoghsgopadi/wav2vec2-large-xlsr-kn")
88
model.to("cuda")
89
90
chars_to_ignore_regex = '[\,\?\.\!\-\;\:\"\“\%\‘\”\�\–\…]'
91
resampler = torchaudio.transforms.Resample(48_000, 16_000)
92
93
# Preprocessing the datasets.
94
# We need to read the aduio files as arrays
95
def speech_file_to_array_fn(batch):
96
    batch["sentence"] = re.sub(chars_to_ignore_regex, '', batch["sentence"]).lower()
97
    speech_array, sampling_rate = torchaudio.load(batch["path"])
98
    batch["speech"] = resampler(speech_array).squeeze().numpy()
99
    return batch
100
101
test_dataset = test_dataset.map(speech_file_to_array_fn)
102
103
# Preprocessing the datasets.
104
# We need to read the aduio files as arrays
105
def evaluate(batch):
106
    inputs = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)
107
    with torch.no_grad():
108
        logits = model(inputs.input_values.to("cuda"), 
109
        attention_mask=inputs.attention_mask.to("cuda")).logits
110
        pred_ids = torch.argmax(logits, dim=-1)
111
        batch["pred_strings"] = processor.batch_decode(pred_ids)
112
        return batch
113
114
result = test_dataset.map(evaluate, batched=True, batch_size=8)
115
print("WER: {:2f}".format(100 * wer.compute(predictions=result["pred_strings"], references=result["sentence"])))
116
117
```
118
119
**Test Result**: 27.08 %  
120
121
## Training
122
123
90% of the OpenSLR Kannada dataset was used for training.
124
125
The colab notebook used for training can be found [here](https://colab.research.google.com/github/amoghgopadi/wav2vec2-xlsr-kannada/blob/main/Fine_Tune_XLSR_Wav2Vec2_on_Kannada_ASR.ipynb).