Wav2vec2 Large Xlsr Japanese Hiragana

🚀 Wav2Vec2-Large-XLSR-53-Japanese

このモデルは、facebook/wav2vec2-large-xlsr-53 を Common Voice と東京大学サルワタリ研究室の日本語音声コーパス JSUT を用いて日本語に対してファインチューニングしたものです。このモデルを使用する際には、音声入力が16kHzでサンプリングされていることを確認してください。

🚀 クイックスタート

モデルの使用

このモデルは、（言語モデルを使用せずに）直接以下のように使用できます。

基本的な使用法

!pip install mecab-python3
!pip install unidic-lite
!pip install pykakasi
!python -m unidic download
import torch
import torchaudio
import librosa
from datasets import load_dataset
import MeCab
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
import re
# config
wakati = MeCab.Tagger("-Owakati")
chars_to_ignore_regex = '[\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\,\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\、\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\。\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\．\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\「\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\」\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\…\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\？\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\・]'
kakasi = pykakasi.kakasi()
kakasi.setMode("J","H")
kakasi.setMode("K","H")
kakasi.setMode("r","Hepburn")
conv = kakasi.getConverter()
# load data, processor and model
test_dataset = load_dataset("common_voice", "ja", split="test[:2%]")
processor = Wav2Vec2Processor.from_pretrained("vumichien/wav2vec2-large-xlsr-japanese-hỉragana")
model = Wav2Vec2ForCTC.from_pretrained("vumichien/wav2vec2-large-xlsr-japanese-hỉragana")
resampler = lambda sr, y: librosa.resample(y.numpy().squeeze(), sr, 16_000)
# Preprocessing the datasets.
def speech_file_to_array_fn(batch):
    batch["sentence"] = conv.do(wakati.parse(batch["sentence"]).strip())
    batch["sentence"] = re.sub(chars_to_ignore_regex,'', batch["sentence"]).strip()
    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)
inputs = processor(test_dataset["speech"][:2], 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)
print("Prediction:", processor.batch_decode(predicted_ids))
print("Reference:", test_dataset["sentence"][:2])

高度な使用法

このモデルは、Common Voiceの日本語テストデータで以下のように評価できます。

!pip install mecab-python3
!pip install unidic-lite
!pip install pykakasi
!python -m unidic download
import torch
import librosa
import torchaudio
from datasets import load_dataset, load_metric
import MeCab
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
import re
#config
wakati = MeCab.Tagger("-Owakati")
chars_to_ignore_regex = '[\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\,\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\、\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\。\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\．\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\「\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\」\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\…\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\？\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\・]'
kakasi = pykakasi.kakasi()
kakasi.setMode("J","H")
kakasi.setMode("K","H")
kakasi.setMode("r","Hepburn")
conv = kakasi.getConverter()
# load data, processor and model
test_dataset = load_dataset("common_voice", "ja", split="test")
wer = load_metric("wer")
cer = load_metric("cer")
processor = Wav2Vec2Processor.from_pretrained("vumichien/wav2vec2-large-xlsr-japanese-hỉragana")
model = Wav2Vec2ForCTC.from_pretrained("vumichien/wav2vec2-large-xlsr-japanese-hỉragana")
model.to("cuda")
resampler = lambda sr, y: librosa.resample(y.numpy().squeeze(), sr, 16_000)
# Preprocessing the datasets.
def speech_file_to_array_fn(batch):
    batch["sentence"] = conv.do(wakati.parse(batch["sentence"]).strip())
    batch["sentence"] = re.sub(chars_to_ignore_regex,'', batch["sentence"]).strip()
    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)
# evaluate function
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"])))
print("CER: {:2f}".format(100 * cer.compute(predictions=result["pred_strings"], references=result["sentence"])))

📚 ドキュメント

テスト結果

• WER: 24.74%
• CER: 10.99%

学習に使用したデータセット

学習には、Common Voiceの train、validation データセットと日本語音声コーパスデータセットが使用されました。

📄 ライセンス

このモデルは、Apache 2.0ライセンスの下で提供されています。