File size: 4,732 Bytes
9c1bbd1 515e217 9c1bbd1 515e217 9c1bbd1 515e217 9c1bbd1 515e217 9c1bbd1 ac53881 515e217 9c1bbd1 515e217 9c1bbd1 515e217 ac53881 515e217 ac53881 515e217 9c1bbd1 515e217 9c1bbd1 515e217 9c1bbd1 515e217 9c1bbd1 515e217 9c1bbd1 515e217 ac53881 515e217 ac53881 515e217 ac53881 515e217 ac53881 515e217 ac53881 515e217 9c1bbd1 515e217 9c1bbd1 515e217 9c1bbd1 |
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 |
---
language: ne
datasets:
- OpenSLR
- common_voice
metrics:
- wer
tags:
- audio
- automatic-speech-recognition
- speech
- xlsr-fine-tuning-week
license: apache-2.0
model-index:
- name: wav2vec2-xlsr-nepali
results:
- task:
name: Speech Recognition
type: automatic-speech-recognition
dataset:
name: OpenSLR ne
type: OpenSLR
args: ne
metrics:
- name: Test WER
type: wer
value: 05.970952
---
# Wav2Vec2-Large-XLSR-53-Nepali
Fine-tuned [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on Nepali using the [Common Voice](https://huggingface.co/datasets/common_voice), and [OpenSLR ne](http://www.openslr.org/43/).
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 torch
import torchaudio
from datasets import load_dataset
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
test_dataset = load_dataset('csv', data_files='/content/ne_np_female/line_index_test.csv',split = 'test')
processor = Wav2Vec2Processor.from_pretrained("gagan3012/wav2vec2-xlsr-nepali")
model = Wav2Vec2ForCTC.from_pretrained("gagan3012/wav2vec2-xlsr-nepali")
resampler = torchaudio.transforms.Resample(48_000, 16_000)
# Preprocessing the datasets.
# We need to read the aduio files as arrays
def speech_file_to_array_fn(batch):
\tspeech_array, sampling_rate = torchaudio.load(batch["path"])
\tbatch["speech"] = resampler(speech_array).squeeze().numpy()
\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])
```
#### Result
Prediction: ['पारानाको ब्राजिली राज्यमा रहेको राजधानी', 'देवराज जोशी त्रिभुवन विश्वविद्यालयबाट शिक्षाशास्त्रमा स्नातक हुनुहुन्छ']
Reference: ['पारानाको ब्राजिली राज्यमा रहेको राजधानी', 'देवराज जोशी त्रिभुवन विश्वविद्यालयबाट शिक्षाशास्त्रमा स्नातक हुनुहुन्छ']
## Evaluation
The model can be evaluated as follows on the {language} test data of Common Voice. # TODO: replace #TODO: replace language with your {language}, *e.g.* French
```python
import torch
import torchaudio
from datasets import load_dataset, load_metric
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
import re
test_dataset = load_dataset('csv', data_files='/content/ne_np_female/line_index_test.csv',split = 'test')
wer = load_metric("wer")
processor = Wav2Vec2Processor.from_pretrained("gagan3012/wav2vec2-xlsr-nepali")
model = Wav2Vec2ForCTC.from_pretrained("gagan3012/wav2vec2-xlsr-nepali")
model.to("cuda")
chars_to_ignore_regex = '[\\,\\?\\.\\!\\-\\;\\:\\"\\“]' # TODO: adapt this list to include all special characters you removed from the data
resampler = torchaudio.transforms.Resample(48_000, 16_000)
# Preprocessing the datasets.
# We need to read the aduio files as arrays
def speech_file_to_array_fn(batch):
\tbatch["sentence"] = re.sub(chars_to_ignore_regex, '', batch["sentence"]).lower()
\tspeech_array, sampling_rate = torchaudio.load(batch["path"])
\tbatch["speech"] = resampler(speech_array).squeeze().numpy()
\treturn 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):
\tinputs = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)
\twith torch.no_grad():
\t\tlogits = model(inputs.input_values.to("cuda"), attention_mask=inputs.attention_mask.to("cuda")).logits
\tpred_ids = torch.argmax(logits, dim=-1)
\tbatch["pred_strings"] = processor.batch_decode(pred_ids)
\treturn 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**: 5.970952 %
## Training
The script used for training can be found [here](https://colab.research.google.com/drive/1AHnYWXb5cwfMEa2o4O3TSdasAR3iVBFP?usp=sharing) |