百分百复现。克隆有电音机械音

[laishujie]

越复用CosyVoice2，电音严重，模型已经更新到最新。

   cosyvoice = CosyVoice2('pretrained_models/CosyVoice2-0.5B', load_jit=False, load_trt=False, fp16=False)

    # NOTE if you want to reproduce the results on https://funaudiollm.github.io/cosyvoice2, please add text_frontend=False during inference
    # zero_shot usage
    prompt_speech_16k = load_wav('test_clone.wav', 16000)
    for i, j in enumerate(cosyvoice.inference_zero_shot(
            '欢迎来到每日AI播客！今天咱们要聊个重磅话题——中国机器人产业如何弯道超车',
            '今天早晨，市中心的主要道路因突发事故造成了严重堵塞，请驾驶员朋友们注意绕行，并听从现场交警的指挥。', prompt_speech_16k, stream=False)):
        torchaudio.save('zero_shot_0.wav'.format(i), j['tts_speech'], cosyvoice.sample_rate)


    prompt_speech_16k = load_wav('test_clone.wav', 16000)
    for i, j in enumerate(cosyvoice.inference_zero_shot(
            '欢迎来到每日AI播客！今天咱们要聊个重磅话题——中国机器人产业如何弯道超车11',
            '今天早晨，市中心的主要道路因突发事故造成了严重堵塞，请驾驶员朋友们注意绕行，并听从现场交警的指挥。', prompt_speech_16k, stream=False)):
        torchaudio.save('zero_shot_1.wav'.format(i), j['tts_speech'], cosyvoice.sample_rate)

克隆素材音频以及结果

clone_and_result.zip