Sew D Tiny 100k Ft Ls100h

🚀 SEW-D-tiny

SEW-D-tiny是基於16kHz採樣語音音頻預訓練的基礎模型。該模型可用於自動語音識別、說話人識別、意圖分類、情感識別等下游任務。使用時，請確保輸入的語音也採樣為16kHz。

🔍 模型信息

📚 相關鏈接

• SEW-D by ASAPP Research
• Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition

📖 論文信息

• 標題：Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition
• 作者：Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi
• 摘要：本文研究了自動語音識別（ASR）預訓練模型中的性能 - 效率權衡問題。聚焦於wav2vec 2.0，本文提出了幾種影響模型性能和效率的架構設計。綜合所有觀察結果，引入了SEW（Squeezed and Efficient Wav2vec），這是一種在各種訓練設置下，在性能和效率方面都有顯著改進的預訓練模型架構。例如，在LibriSpeech的100h - 960h半監督設置下，與wav2vec 2.0相比，SEW的推理速度提高了1.9倍，單詞錯誤率相對降低了13.5%。在相似的推理時間內，SEW在不同模型大小下將單詞錯誤率降低了25 - 50%。
• 原始模型：https://github.com/asappresearch/sew#model-checkpoints

🚀 快速開始

💻 使用示例

基礎用法

from transformers import Wav2Vec2Processor, SEWDForCTC
from datasets import load_dataset
import soundfile as sf
import torch
 
# load the model and preprocessor
processor = Wav2Vec2Processor.from_pretrained("asapp/sew-d-tiny-100k-ft-ls100h")
model = SEWDForCTC.from_pretrained("asapp/sew-d-tiny-100k-ft-ls100h")

# load the dummy dataset with speech samples
ds = load_dataset("patrickvonplaten/librispeech_asr_dummy", "clean", split="validation")
 
# preprocess
input_values = processor(ds[0]["audio"]["array"], return_tensors="pt").input_values  # Batch size 1

# retrieve logits
logits = model(input_values).logits
 
# take argmax and decode
predicted_ids = torch.argmax(logits, dim=-1)
transcription = processor.batch_decode(predicted_ids)

評估示例

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

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

model = SEWDForCTC.from_pretrained("asapp/sew-d-tiny-100k-ft-ls100h").to("cuda")
processor = Wav2Vec2Processor.from_pretrained("asapp/sew-d-tiny-100k-ft-ls100h")

def map_to_pred(batch):
    input_values = processor(batch["audio"][0]["array"], sampling_rate=16000, 
                             return_tensors="pt", padding="longest").input_values
    with torch.no_grad():
        logits = model(input_values.to("cuda")).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, batched=True, batch_size=1, remove_columns=["audio"])

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

📊 評估結果

📄 許可證

本項目採用Apache-2.0許可證。