标签:

• 训练生成 模型索引:
• 名称: wav2vec2-russian 结果: [] 小部件:
• 示例: https://cdn-media.huggingface.co/speech_samples/common_voice_ru_18849022.mp3

wav2vec2-russian

用于音频识别的模型。模型输出结果可进一步通过我的文本纠错网络UrukHan/t5-russian-spell进行处理。

请上传wav格式的音频文件进行识别。识别结果可通过我的另一网络进行修正，该网络能纠正错误、添加标点并修正数字格式。访问地址：https://huggingface.co/UrukHan/t5-russian-spell

运行网络 示例参见Colab笔记本 https://colab.research.google.com/drive/1dVZvccYJq02hmEsapWgmuJ-pLdezFnn1?usp=sharing

from transformers import AutoModelForCTC, Wav2Vec2Processor
model = AutoModelForCTC.from_pretrained("UrukHan/wav2vec2-russian")
processor = Wav2Vec2Processor.from_pretrained("UrukHan/wav2vec2-russian")
def map_to_result(batch):
  with torch.no_grad():
    input_values = torch.tensor(batch["input_values"]).unsqueeze(0) #, device="cuda"
    logits = model(input_values).logits
  pred_ids = torch.argmax(logits, dim=-1)
  batch = processor.batch_decode(pred_ids)[0]
  return batch
 
 map_to_result()

模型训练与数据处理及数据集创建详解请参考Colab笔记本：

https://colab.research.google.com/drive/1zkCA2PtKxD2acqLr55USh35OomoOwOhm?usp=sharing