Wav2vec2 Large Xlsr Icelandic

🚀 Wav2Vec2-Large-XLSR-53-Icelandic

このモデルは、Malromurを使用してアイスランド語で微調整されたfacebook/wav2vec2-large-xlsr-53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

このモデルを使用するには、以下の手順に従ってください。

✨ 主な機能

• アイスランド語の自動音声認識に特化したモデルです。
• 微調整により、アイスランド語の音声認識精度が向上しています。

📦 インストール

このモデルを使用するために必要なパッケージをインストールします。

必要なパッケージのインストール

# requirement packages
!pip install git+https://github.com/huggingface/datasets.git
!pip install git+https://github.com/huggingface/transformers.git
!pip install torchaudio
!pip install librosa
!pip install jiwer
!pip install num2words

数値変換パッケージのダウンロード

# num2word packages
# Original source: https://github.com/savoirfairelinux/num2words
!mkdir -p ./num2words
!wget -O num2words/__init__.py https://huggingface.co/m3hrdadfi/wav2vec2-large-xlsr-icelandic/raw/main/num2words/__init__.py
!wget -O num2words/base.py https://huggingface.co/m3hrdadfi/wav2vec2-large-xlsr-icelandic/raw/main/num2words/base.py
!wget -O num2words/compat.py https://huggingface.co/m3hrdadfi/wav2vec2-large-xlsr-icelandic/raw/main/num2words/compat.py
!wget -O num2words/currency.py https://huggingface.co/m3hrdadfi/wav2vec2-large-xlsr-icelandic/raw/main/num2words/currency.py
!wget -O num2words/lang_EU.py https://huggingface.co/m3hrdadfi/wav2vec2-large-xlsr-icelandic/raw/main/num2words/lang_EU.py
!wget -O num2words/lang_IS.py https://huggingface.co/m3hrdadfi/wav2vec2-large-xlsr-icelandic/raw/main/num2words/lang_IS.py
!wget -O num2words/utils.py https://huggingface.co/m3hrdadfi/wav2vec2-large-xlsr-icelandic/raw/main/num2words/utils.py

# Malromur_test selected based on gender and age
!wget -O malromur_test.csv https://huggingface.co/m3hrdadfi/wav2vec2-large-xlsr-icelandic/raw/main/malromur_test.csv

# Normalizer
!wget -O normalizer.py https://huggingface.co/m3hrdadfi/wav2vec2-large-xlsr-icelandic/raw/main/normalizer.py

💻 使用例

基本的な使用法

import librosa
import torch
import torchaudio
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
from datasets import load_dataset

import numpy as np
import re
import string

import IPython.display as ipd

from normalizer import Normalizer

normalizer = Normalizer(lang="is")


def speech_file_to_array_fn(batch):
    speech_array, sampling_rate = torchaudio.load(batch["path"])
    speech_array = speech_array.squeeze().numpy()
    speech_array = librosa.resample(np.asarray(speech_array), sampling_rate, 16_000)

    batch["speech"] = speech_array
    return batch


def predict(batch):
    features = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)

    input_values = features.input_values.to(device)
    attention_mask = features.attention_mask.to(device)

    with torch.no_grad():
        logits = model(input_values, attention_mask=attention_mask).logits 
        
    pred_ids = torch.argmax(logits, dim=-1)

    batch["predicted"] = processor.batch_decode(pred_ids)
    return batch


device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
processor = Wav2Vec2Processor.from_pretrained("m3hrdadfi/wav2vec2-large-xlsr-icelandic")
model = Wav2Vec2ForCTC.from_pretrained("m3hrdadfi/wav2vec2-large-xlsr-icelandic").to(device)

dataset = load_dataset("csv", data_files={"test": "./malromur_test.csv"})["test"]
dataset = dataset.map(
    normalizer, 
    fn_kwargs={"do_lastspace_removing": True, "text_key_name": "cleaned_sentence"},
    remove_columns=list(set(dataset.column_names) - set(['cleaned_sentence', 'path']))
)

dataset = dataset.map(speech_file_to_array_fn)
result = dataset.map(predict, batched=True, batch_size=8)

max_items = np.random.randint(0, len(result), 20).tolist()
for i in max_items:
    reference, predicted =  result["cleaned_sentence"][i], result["predicted"][i]
    print("reference:", reference)
    print("predicted:", predicted)
    print('---')

出力例

reference: eða eitthvað annað dýr
predicted: eða eitthvað annað dýr
---
reference: oddgerður
predicted: oddgerður
---
reference: eiðný
predicted: eiðný
---
reference: löndum
predicted: löndum
---
reference: tileinkaði bróður sínum markið
predicted: tileinkaði bróður sínum markið
---
reference: þetta er svo mikill hégómi
predicted: þetta er svo mikill hégómi
---
reference: timarit is
predicted: timarit is
---
reference: stefna strax upp aftur
predicted: stefna strax upp aftur
---
reference: brekkuflöt
predicted: brekkuflöt
---
reference: áætlunarferð frestað vegna veðurs
predicted: áætluna ferð frestað vegna veðurs
---
reference: sagði af sér vegna kláms
predicted: sagði af sér vegni kláms
---
reference: grímúlfur
predicted: grímúlgur
---
reference: lýsti sig saklausan
predicted: lýsti sig saklausan
---
reference: belgingur is
predicted: belgingur is
---
reference: sambía
predicted: sambía
---
reference: geirastöðum
predicted: geirastöðum
---
reference: varð tvisvar fyrir eigin bíl
predicted: var tvisvar fyrir eigin bíl
---
reference: reykjavöllum
predicted: reykjavöllum
---
reference: miklir menn eru þeir þremenningar
predicted: miklir menn eru þeir þremenningar
---
reference: handverkoghonnun is
predicted: handverkoghonnun is
---

モデルの評価

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

import numpy as np
import re
import string

from normalizer import Normalizer

normalizer = Normalizer(lang="is")


def speech_file_to_array_fn(batch):
    speech_array, sampling_rate = torchaudio.load(batch["path"])
    speech_array = speech_array.squeeze().numpy()
    speech_array = librosa.resample(np.asarray(speech_array), sampling_rate, 16_000)

    batch["speech"] = speech_array
    return batch


def predict(batch):
    features = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)

    input_values = features.input_values.to(device)
    attention_mask = features.attention_mask.to(device)

    with torch.no_grad():
        logits = model(input_values, attention_mask=attention_mask).logits 
        
    pred_ids = torch.argmax(logits, dim=-1)

    batch["predicted"] = processor.batch_decode(pred_ids)
    return batch


device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
processor = Wav2Vec2Processor.from_pretrained("m3hrdadfi/wav2vec2-large-xlsr-icelandic")
model = Wav2Vec2ForCTC.from_pretrained("m3hrdadfi/wav2vec2-large-xlsr-icelandic").to(device)

dataset = load_dataset("csv", data_files={"test": "./malromur_test.csv"})["test"]
dataset = dataset.map(
    normalizer, 
    fn_kwargs={"do_lastspace_removing": True, "text_key_name": "cleaned_sentence"},
    remove_columns=list(set(dataset.column_names) - set(['cleaned_sentence', 'path']))
)

dataset = dataset.map(speech_file_to_array_fn)
result = dataset.map(predict, batched=True, batch_size=8)

wer = load_metric("wer")

print("WER: {:.2f}".format(100 * wer.compute(predictions=result["predicted"], references=result["cleaned_sentence"])))

テスト結果:

• WER: 09.21%

📚 ドキュメント

学習について

Common Voiceのtrainとvalidationデータセットを学習に使用しました。

学習状況はこちらで確認できます。

学習に使用したスクリプトはこちらにあります。

質問や問題の報告

Wav2VecのリポジトリにGitHubのissueを投稿してください。

📄 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。