Wav2vec2 Conformer Rel Pos Large 960h Ft

🚀 Wav2Vec2-Conformer-Large-960h 带相对位置嵌入

Wav2Vec2-Conformer 带有相对位置嵌入，在16kHz采样的语音音频上进行了预训练，并在960小时的Librispeech数据集上进行了微调。使用该模型时，请确保您的语音输入也是16kHz采样的。

📚 详细文档

• 论文：fairseq S2T: Fast Speech-to-Text Modeling with fairseq
• 作者：Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, Juan Pino

Wav2Vec2-Conformer的实验结果可在官方论文的表3和表4中找到。

原始模型可在https://github.com/pytorch/fairseq/tree/master/examples/wav2vec#wav2vec-20 找到。

📦 模型信息

📊 评估结果

💻 使用示例

基础用法

from transformers import Wav2Vec2Processor, Wav2Vec2ConformerForCTC
from datasets import load_dataset
import torch

# 加载模型和处理器
processor = Wav2Vec2Processor.from_pretrained("facebook/wav2vec2-conformer-rel-pos-large-960h-ft")
model = Wav2Vec2ConformerForCTC.from_pretrained("facebook/wav2vec2-conformer-rel-pos-large-960h-ft")

# 加载示例数据集并读取音频文件
ds = load_dataset("patrickvonplaten/librispeech_asr_dummy", "clean", split="validation")

# 进行分词
input_values = processor(ds[0]["audio"]["array"], return_tensors="pt", padding="longest").input_values

# 获取对数概率
logits = model(input_values).logits

# 取最大值并解码
predicted_ids = torch.argmax(logits, dim=-1)
transcription = processor.batch_decode(predicted_ids)

评估示例

此代码片段展示了如何在LibriSpeech的“clean”和“other”测试数据上评估facebook/wav2vec2-conformer-rel-pos-large-960h-ft。

from datasets import load_dataset
from transformers import Wav2Vec2ConformerForCTC, Wav2Vec2Processor
import torch
from jiwer import wer

librispeech_eval = load_dataset("librispeech_asr", "clean", split="test")

model = Wav2Vec2ConformerForCTC.from_pretrained("facebook/wav2vec2-large-960h-lv60-self").to("cuda")
processor = Wav2Vec2Processor.from_pretrained("facebook/wav2vec2-large-960h-lv60-self")

def map_to_pred(batch):
    inputs = processor(batch["audio"]["array"], return_tensors="pt", padding="longest")
    input_values = inputs.input_values.to("cuda")
    attention_mask = inputs.attention_mask.to("cuda")

    with torch.no_grad():
        logits = model(input_values, attention_mask=attention_mask).logits

    predicted_ids = torch.argmax(logits, dim=-1)
    transcription = processor.batch_decode(predicted_ids)
    batch["transcription"] = transcription
    return batch

result = librispeech_eval.map(map_to_pred, remove_columns=["audio"])

print("WER:", wer(result["text"], result["transcription"]))

结果 (字错率 WER):