Wav2vec2 Large Xlsr Icelandic

🚀 Wav2Vec2-Large-XLSR-53-冰島語模型

本項目基於 Malromur 數據集，對 facebook/wav2vec2-large-xlsr-53 模型進行了冰島語微調。使用該模型時，請確保輸入的語音採樣率為 16kHz。

🚀 快速開始

本模型可直接使用（無需語言模型），以下是詳細的使用步驟。

✨ 主要特性

• 語言適配：針對冰島語進行了微調，能更好地處理冰島語語音識別任務。
• 多數據集支持：使用了 Malromur 數據集進行訓練和評估。

📦 安裝指南

安裝依賴包

# 安裝所需的包
!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

下載 num2word 相關文件

# 創建 num2words 目錄
!mkdir -p ./num2words
# 下載 num2words 相關文件
!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

# 下載 Malromur 測試集
!wget -O malromur_test.csv https://huggingface.co/m3hrdadfi/wav2vec2-large-xlsr-icelandic/raw/main/malromur_test.csv

# 下載歸一化器
!wget -O normalizer.py https://huggingface.co/m3hrdadfi/wav2vec2-large-xlsr-icelandic/raw/main/normalizer.py

💻 使用示例

基礎用法

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('---')

輸出示例

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
---

📚 詳細文檔

評估模型

可以使用以下代碼在 Malromur 測試集上評估模型：

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"])))

測試結果

• 字錯率（WER）: 09.21%

訓練與報告

• 訓練數據集：使用了 Common Voice 的 train 和 validation 數據集進行訓練。
• 訓練狀態：可以在 這裡 查看訓練狀態。
• 訓練腳本：訓練使用的腳本可以在 這裡 找到。

常見問題

如果有任何問題，請在 Wav2Vec 倉庫中提交 GitHub issue。

📄 許可證

本項目採用 Apache-2.0 許可證。

📋 模型信息