语言: 罗曼什语-瓦拉迪尔方言
数据集:

• 通用语音(common_voice)
  评估指标:
• 词错误率(wer)
  标签:
• 音频
• 自动语音识别
• 语音
• XLSR微调周
  许可证: Apache-2.0
  模型索引:
• 名称: Anurag Singh XLSR Wav2Vec2 Large 53 罗曼什语瓦拉迪尔方言
  结果:
  • 任务:
    名称: 语音识别
    类型: 自动语音识别
    数据集:
    名称: 通用语音 rm-vallader
    类型: common_voice
    参数: rm-vallader
    指标:
    • 名称: 测试词错误率
      类型: wer
      值: 32.89

Wav2Vec2-Large-XLSR-53-罗曼什语瓦拉迪尔方言

基于facebook/wav2vec2-large-xlsr-53模型，使用通用语音数据集对罗曼什语瓦拉迪尔方言进行微调。
使用该模型时，请确保语音输入采样率为16kHz。

使用方法

该模型可直接使用（无需语言模型）如下：

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

test_dataset = load_dataset("common_voice", "rm-vallader", split="test[:2%]")  
processor = Wav2Vec2Processor.from_pretrained("anuragshas/wav2vec2-large-xlsr-53-rm-vallader")  
model = Wav2Vec2ForCTC.from_pretrained("anuragshas/wav2vec2-large-xlsr-53-rm-vallader")  
resampler = torchaudio.transforms.Resample(48_000, 16_000)  

# 数据预处理  
# 需要将音频文件读取为数组  
def speech_file_to_array_fn(batch):  
    speech_array, sampling_rate = torchaudio.load(batch["path"])  
    batch["speech"] = resampler(speech_array).squeeze().numpy()  
    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("预测结果:", processor.batch_decode(predicted_ids))  
print("参考文本:", test_dataset["sentence"][:2])  

评估

可在通用语音数据集的罗曼什语瓦拉迪尔方言测试集上评估模型如下：

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

test_dataset = load_dataset("common_voice", "rm-vallader", split="test")  
wer = load_metric("wer")  
processor = Wav2Vec2Processor.from_pretrained("anuragshas/wav2vec2-large-xlsr-53-rm-vallader")  
model = Wav2Vec2ForCTC.from_pretrained("anuragshas/wav2vec2-large-xlsr-53-rm-vallader")  
model.to("cuda")  

chars_to_ignore_regex = '[\,\?\.\!\-\;\:\"\“\%\”\„\–\…\«\»]'  
resampler = torchaudio.transforms.Resample(48_000, 16_000)  

# 数据预处理  
def speech_file_to_array_fn(batch):  
    batch["sentence"] = re.sub('’ ',' ',batch["sentence"])  
    batch["sentence"] = re.sub(' ‘',' ',batch["sentence"])  
    batch["sentence"] = re.sub('’|‘','\'',batch["sentence"])  
    batch["sentence"] = re.sub(chars_to_ignore_regex, '', batch["sentence"]).lower()  
    speech_array, sampling_rate = torchaudio.load(batch["path"])  
    batch["speech"] = resampler(speech_array).squeeze().numpy()  
    return batch  

test_dataset = test_dataset.map(speech_file_to_array_fn)  

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("词错误率: {:2f}".format(100 * wer.compute(predictions=result["pred_strings"], references=result["sentence"])))  

测试结果: 32.89%

训练过程

使用通用语音数据集的train和validation子集进行训练。