About missing .npy file if you have uploaded .wav file (如果上传了音频但没找到对应.npy文件请看这里)

[jiang-ziyue]

• Format Requirements: All audio files must be in .wav format and shorter than 24 seconds. Filenames must not contain spaces. Files that do not meet these requirements will be deleted.

• If the uploaded audio file is in the correct format but the corresponding .npy file is still missing after 1-2 days, you can re-upload the audio file and reply with its filename below. We’ll process it for you as soon as possible.

• 格式要求： 所有音频必须是.wav格式且时长小于24秒，文件名中不能包含空格。不符合要求的文件会被删除。

• 如果上传的音频符合文件格式但仍然在1～2天后找不到对应的.npy文件，您可以重新上传一次音频文件，并在下方回复您的文件名。我们会尽快为您处理

[baifachuan]

(MegaTTS3) fcbai@fcbai-w5:~/workspace/code/MegaTTS3$ python tts/infer_cli.py --input_wav '/home/fcbai/Downloads/format_denoise_20250411204855084.wav'  --input_text "另一边的桌上,一位读书人嗤之以鼻道,'佛子三藏,神子燕小鱼是什么样的人物,李家的那个李子夜如何与他们相提并论？'" --output_dir ./gen
| loaded 'dur_model' from './checkpoints/duration_lm/model_only_last.ckpt'.
| Missing keys: 0, Unexpected keys: 0
| loaded 'dit' from './checkpoints/diffusion_transformer/model_only_last.ckpt'.
| Missing keys: 0, Unexpected keys: 9
| loaded 'model' from './checkpoints/aligner_lm/model_only_last.ckpt'.
| Missing keys: 0, Unexpected keys: 0
/home/fcbai/.virtualenvs/MegaTTS3/lib/python3.10/site-packages/torch/nn/utils/weight_norm.py:28: UserWarning: torch.nn.utils.weight_norm is deprecated in favor of torch.nn.utils.parametrizations.weight_norm.
  warnings.warn("torch.nn.utils.weight_norm is deprecated in favor of torch.nn.utils.parametrizations.weight_norm.")
| loaded 'model_gen' from './checkpoints/wavvae/decoder.ckpt'.
| Missing keys: 74, Unexpected keys: 0
2025-04-13 14:34:58,381 WETEXT INFO found existing fst: /home/fcbai/.virtualenvs/MegaTTS3/lib/python3.10/site-packages/tn/zh_tn_tagger.fst
2025-04-13 14:34:58,381 WETEXT INFO                     /home/fcbai/.virtualenvs/MegaTTS3/lib/python3.10/site-packages/tn/zh_tn_verbalizer.fst
2025-04-13 14:34:58,381 WETEXT INFO skip building fst for zh_normalizer ...
2025-04-13 14:34:58,521 WETEXT INFO found existing fst: /home/fcbai/.virtualenvs/MegaTTS3/lib/python3.10/site-packages/tn/en_tn_tagger.fst
2025-04-13 14:34:58,521 WETEXT INFO                     /home/fcbai/.virtualenvs/MegaTTS3/lib/python3.10/site-packages/tn/en_tn_verbalizer.fst
2025-04-13 14:34:58,521 WETEXT INFO skip building fst for en_normalizer ...
| Start processing /home/fcbai/Downloads/format_denoise_20250411204855084.wav+另一边的桌上,一位读书人嗤之以鼻道,'佛子三藏,神子燕小鱼是什么样的人物,李家的那个李子夜如何与他们相提并论？'
Traceback (most recent call last):
  File "/home/fcbai/workspace/code/MegaTTS3/tts/infer_cli.py", line 273, in <module>
    resource_context = infer_ins.preprocess(file_content, latent_file=wav_path.replace('.wav', '.npy'))
  File "/home/fcbai/workspace/code/MegaTTS3/tts/infer_cli.py", line 186, in preprocess
    vae_latent = torch.from_numpy(np.load(latent_file)).to(self.device)
  File "/home/fcbai/.virtualenvs/MegaTTS3/lib/python3.10/site-packages/numpy/lib/npyio.py", line 427, in load
    fid = stack.enter_context(open(os_fspath(file), "rb"))
FileNotFoundError: [Errno 2] No such file or directory: '/home/fcbai/Downloads/format_denoise_20250411204855084.npy'

[jiang-ziyue]

@baifachuan hey baifachuan, did you upload format_denoise_20250411204855084.wav to our google drive? Our project requires you to upload the .wav file to the google drive to obtain the format_denoise_20250411204855084.npy

[miovovo]

@jiang-ziyue 我想要测试seed_test数据集，我已经上传了seed_en.tar和seed_zh.tar两个压缩包，需要多长时间可以得到npy文件？

[jiang-ziyue]

@miovovo 您好，这周五之前，等我们传好后告诉您对应的链接。

[Tybost]

@jiang-ziyue MegaTTS3 is pretty good but will we be able to train our own .NPY locally in the future? I would assume you aren't going to endlessly manually train .NPY files. Thank you and great work!

[lhybdv]

@jiang-ziyue
我上传了 主持人配音A0.wav，请帮忙生成一下 npy 文件, 感谢！

[AWMANW]

我前天上传了Sample-woman1.wav和Sample-woman2.wav，请帮忙生成一下 npy 文件, 感谢！

[vdruts]

Will you ever release ability to create our own files?

[GanziPo]

@jiang-ziyue 我上传了 hg.mp3和pq.wav，请帮忙生成一下 npy 文件, 感谢！

[Cat-Dolly]

您好，我上传了“智小柔.wav"和“飞镜.wav“，请帮忙生成一下npy文件，感谢～

[Z-yq]

也可以尝试 https://modelscope.cn/models/ACoderPassBy/MegaTTS-SFT 这套非官方的模型参数

[muzesunshine]

我上传了音频文件yueli.wav ,请帮忙生成npy文件，谢谢~ @jiang-ziyue

[booksword]

6.2上传的wav文件丹青帮忙生成npy文件，谢谢

tmk陈璐.wav
tmk朱娟.wav
tmk陈珍彩.wav
tmk周琴.wav

[rogerhanxy]

6月4日上传的wav文件帮忙生成npy文件，谢谢
rogerhanxy.wav
rogerhan2.wav