File size: 4,567 Bytes
a69148f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
language: hi
#datasets:
#- Interspeech 2021
metrics:
- wer
tags:
- audio
- automatic-speech-recognition
- speech
license: mit
model-index:
- name: Wav2Vec2 Vakyansh Hindi Model by Harveen Chadha
  results:
  - task:
      name: Speech Recognition
      type: automatic-speech-recognition
    dataset:
      name: Common Voice hi
      type: common_voice
      args: hi
    metrics:
    - name: Test WER
      type: wer
      value: 33.17
---

## Spaces Demo
Check the spaces demo [here](https://huggingface.co/spaces/Harveenchadha/wav2vec2-vakyansh-hindi/tree/main)

## Pretrained Model

Fine-tuned on Multilingual Pretrained Model [CLSRIL-23](https://arxiv.org/abs/2107.07402). The original fairseq checkpoint is present [here](https://github.com/Open-Speech-EkStep/vakyansh-models). When using this model, make sure that your speech input is sampled at 16kHz.

**Note: The result from this model is without a language model so you may witness a higher WER in some cases.**

## Dataset

This model was trained on 4200 hours of Hindi Labelled Data. The labelled data is not present in public domain as of now.

## Training Script

Models were trained using experimental platform setup by Vakyansh team at Ekstep. Here is the [training repository](https://github.com/Open-Speech-EkStep/vakyansh-wav2vec2-experimentation).

In case you want to explore training logs on wandb they are [here](https://wandb.ai/harveenchadha/hindi_finetuning_multilingual?workspace=user-harveenchadha).


## [Colab Demo](https://colab.research.google.com/github/harveenchadha/bol/blob/main/demos/hf/hindi/hf_hindi_him_4200_demo.ipynb)

## Usage

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

```python
import soundfile as sf
import torch
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
import argparse

def parse_transcription(wav_file):
    # load pretrained model
    processor = Wav2Vec2Processor.from_pretrained("Harveenchadha/vakyansh-wav2vec2-hindi-him-4200")
    model = Wav2Vec2ForCTC.from_pretrained("Harveenchadha/vakyansh-wav2vec2-hindi-him-4200")

    # load audio
    audio_input, sample_rate = sf.read(wav_file)

    # pad input values and return pt tensor
    input_values = processor(audio_input, sampling_rate=sample_rate, return_tensors="pt").input_values

    # INFERENCE
    # retrieve logits & take argmax
    logits = model(input_values).logits
    predicted_ids = torch.argmax(logits, dim=-1)

    # transcribe
    transcription = processor.decode(predicted_ids[0], skip_special_tokens=True)
    print(transcription)

```


## Evaluation
The model can be evaluated as follows on the hindi test data of Common Voice. 

```python

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

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

processor = Wav2Vec2Processor.from_pretrained("Harveenchadha/vakyansh-wav2vec2-hindi-him-4200")
model = Wav2Vec2ForCTC.from_pretrained("Harveenchadha/vakyansh-wav2vec2-hindi-him-4200")
model.to("cuda")

resampler = torchaudio.transforms.Resample(48_000, 16_000)

chars_to_ignore_regex = '[\,\?\.\!\-\;\:\"\“]'

# Preprocessing the datasets.
# We need to read the aduio 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(speech_array).squeeze().numpy()
  return batch

test_dataset = test_dataset.map(speech_file_to_array_fn)

# Preprocessing the datasets.
# We need to read the aduio 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")).logits

      pred_ids = torch.argmax(logits, dim=-1)
      batch["pred_strings"] = processor.batch_decode(pred_ids, skip_special_tokens=True)
      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**: 33.17 %

[**Colab Evaluation**](https://colab.research.google.com/github/harveenchadha/bol/blob/main/demos/hf/hindi/hf_vakyansh_hindi_him_4200_evaluation_common_voice.ipynb) 

## Credits
Thanks to Ekstep Foundation for making this possible. The vakyansh team will be open sourcing speech models in all the Indic Languages.