File size: 4,451 Bytes
02dfab0 a7542f7 02dfab0 a7542f7 02dfab0 00301b7 02dfab0 be8fa13 02dfab0 be8fa13 1b7d8af 3446090 1b7d8af be8fa13 02dfab0 fb1543e 02dfab0 1934b66 7102579 02dfab0 7bbf0ec 7102579 7bbf0ec 02dfab0 bcfe680 fb1543e 5dadb1e bcfe680 3d42281 bcfe680 f7c7884 |
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 |
---
language:
- ru
license: apache-2.0
tags:
- automatic-speech-recognition
- hf-asr-leaderboard
- mozilla-foundation/common_voice_8_0
- robust-speech-event
- ru
datasets:
- mozilla-foundation/common_voice_8_0
model-index:
- name: XLS-R Wav2Vec2 Russian by Jonatas Grosman
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Common Voice 8
type: mozilla-foundation/common_voice_8_0
args: ru
metrics:
- name: Test WER
type: wer
value: 9.82
- name: Test CER
type: cer
value: 2.3
- name: Test WER (+LM)
type: wer
value: 7.08
- name: Test CER (+LM)
type: cer
value: 1.87
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Robust Speech Event - Dev Data
type: speech-recognition-community-v2/dev_data
args: ru
metrics:
- name: Dev WER
type: wer
value: 23.96
- name: Dev CER
type: cer
value: 8.88
- name: Dev WER (+LM)
type: wer
value: 15.88
- name: Dev CER (+LM)
type: cer
value: 7.42
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Robust Speech Event - Test Data
type: speech-recognition-community-v2/eval_data
args: ru
metrics:
- name: Test WER
type: wer
value: 14.23
---
# Fine-tuned XLS-R 1B model for speech recognition in Russian
Fine-tuned [facebook/wav2vec2-xls-r-1b](https://huggingface.co/facebook/wav2vec2-xls-r-1b) on Russian using the train and validation splits of [Common Voice 8.0](https://huggingface.co/datasets/mozilla-foundation/common_voice_8_0), [Golos](https://www.openslr.org/114/), and [Multilingual TEDx](http://www.openslr.org/100).
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, and thanks to the GPU credits generously given by the [OVHcloud](https://www.ovhcloud.com/en/public-cloud/ai-training/) :)
## Usage
Using the [HuggingSound](https://github.com/jonatasgrosman/huggingsound) library:
```python
from huggingsound import SpeechRecognitionModel
model = SpeechRecognitionModel("jonatasgrosman/wav2vec2-xls-r-1b-russian")
audio_paths = ["/path/to/file.mp3", "/path/to/another_file.wav"]
transcriptions = model.transcribe(audio_paths)
```
Writing your own inference script:
```python
import torch
import librosa
from datasets import load_dataset
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
LANG_ID = "ru"
MODEL_ID = "jonatasgrosman/wav2vec2-xls-r-1b-russian"
SAMPLES = 10
test_dataset = load_dataset("common_voice", LANG_ID, split=f"test[:{SAMPLES}]")
processor = Wav2Vec2Processor.from_pretrained(MODEL_ID)
model = Wav2Vec2ForCTC.from_pretrained(MODEL_ID)
# Preprocessing the datasets.
# We need to read the audio files as arrays
def speech_file_to_array_fn(batch):
speech_array, sampling_rate = librosa.load(batch["path"], sr=16_000)
batch["speech"] = speech_array
batch["sentence"] = batch["sentence"].upper()
return batch
test_dataset = test_dataset.map(speech_file_to_array_fn)
inputs = processor(test_dataset["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)
with torch.no_grad():
logits = model(inputs.input_values, attention_mask=inputs.attention_mask).logits
predicted_ids = torch.argmax(logits, dim=-1)
predicted_sentences = processor.batch_decode(predicted_ids)
```
## Evaluation Commands
1. To evaluate on `mozilla-foundation/common_voice_8_0` with split `test`
```bash
python eval.py --model_id jonatasgrosman/wav2vec2-xls-r-1b-russian --dataset mozilla-foundation/common_voice_8_0 --config ru --split test
```
2. To evaluate on `speech-recognition-community-v2/dev_data`
```bash
python eval.py --model_id jonatasgrosman/wav2vec2-xls-r-1b-russian --dataset speech-recognition-community-v2/dev_data --config ru --split validation --chunk_length_s 5.0 --stride_length_s 1.0
```
## Citation
If you want to cite this model you can use this:
```bibtex
@misc{grosman2021xlsr-1b-russian,
title={Fine-tuned {XLS-R} 1{B} model for speech recognition in {R}ussian},
author={Grosman, Jonatas},
howpublished={\url{https://huggingface.co/jonatasgrosman/wav2vec2-xls-r-1b-russian}},
year={2022}
}
``` |