[GH-ISSUE #23] Any docs/examples on how to do the audio separation with HP5_only_main_vocal and DeEcho #20

New issue

Open

opened 2026-10-20 09:53:10 +00:00 by bsakaguchi · 4 comments

bsakaguchi commented

2026-10-20 09:53:10 +00:00

Owner

Originally created by @ghost on GitHub (Mar 20, 2024).
Original GitHub issue: https://github.com/RVC-Project/Retrieval-based-Voice-Conversion/issues/23

Is this available in this version of the project yet? if so any examples, trying to migrate from the old repo where it was working.

thanks,

Originally created by @ghost on GitHub (Mar 20, 2024). Original GitHub issue: https://github.com/RVC-Project/Retrieval-based-Voice-Conversion/issues/23 Is this available in this version of the project yet? if so any examples, trying to migrate from the old repo where it was working. thanks,

bsakaguchi added the

bug

help wanted

labels

2026-10-20 09:53:10 +00:00

bsakaguchi commented

2026-10-20 09:53:11 +00:00

Author

Owner

@Tps-F commented on GitHub (Mar 22, 2024):

I'll make it!

@Tps-F commented on GitHub (Mar 22, 2024): I'll make it!

bsakaguchi commented

2026-10-20 09:53:11 +00:00

Author

Owner

@ghost commented on GitHub (Apr 2, 2024):

doing something like this returns me two files that are the same as the input, what am I getting wrong?

from rvc.modules.uvr5.vr import AudioPreprocess

def separate():
    input_path = f"{project_dir}/sandbox/inputs/input.mp3"
    output_path = Path(f"{project_dir}/sandbox/outputs/")

    model_name = "HP5_only_main_vocal.pth"
    model_path = os.path.join(os.getenv("weight_uvr5_root"), model_name)
    agg = 10  

    pre_fun = AudioPreprocess(model_path, agg)
    result = pre_fun.process(input_path)

    wav_instrument, wav_vocals, sr, agg = result
    sf.write(f"{output_path}/instrumental.wav", wav_instrument / 32768, sr)
    sf.write(f"{output_path}/vocals.wav", wav_vocals / 32768, sr)

@ghost commented on GitHub (Apr 2, 2024): doing something like this returns me two files that are the same as the input, what am I getting wrong? ``` from rvc.modules.uvr5.vr import AudioPreprocess def separate(): input_path = f"{project_dir}/sandbox/inputs/input.mp3" output_path = Path(f"{project_dir}/sandbox/outputs/") model_name = "HP5_only_main_vocal.pth" model_path = os.path.join(os.getenv("weight_uvr5_root"), model_name) agg = 10 pre_fun = AudioPreprocess(model_path, agg) result = pre_fun.process(input_path) wav_instrument, wav_vocals, sr, agg = result sf.write(f"{output_path}/instrumental.wav", wav_instrument / 32768, sr) sf.write(f"{output_path}/vocals.wav", wav_vocals / 32768, sr) ```

bsakaguchi commented

2026-10-20 09:53:11 +00:00

Author

Owner

@ghost commented on GitHub (Apr 9, 2024):

The cli and api endpoint seem to make use of a wrapper:

    uvr_module.uvr_wrapper(
        inputpath, outputpath, model_name=modelname, export_format=format
    )
    ```
    
    But not sure this actually works, as the wrapper does not accept an output_path, it returns the files.
    
    Have also tried to use the wrapper directly:
    
    ```
        uvr = UVR()
    results = uvr.uvr_wrapper(input_path, agg=10, model_name="HP3_all_vocals.pth")

    for i, result in enumerate(results):
        wav_instrument, wav_vocals, sr, agg = result
        sf.write(f"{output_path}/instrument_{i}.wav", wav_instrument / 32768, sr)
        sf.write(f"{output_path}/vocals_{i}.wav", wav_vocals / 32768, sr)

but sadly I still get the same issue of the instrumental and vocals files being the same (original audio). Any help @Tps-F ?

@ghost commented on GitHub (Apr 9, 2024): The cli and api endpoint seem to make use of a wrapper: ``` uvr_module.uvr_wrapper( inputpath, outputpath, model_name=modelname, export_format=format ) ``` But not sure this actually works, as the wrapper does not accept an output_path, it returns the files. Have also tried to use the wrapper directly: ``` uvr = UVR() results = uvr.uvr_wrapper(input_path, agg=10, model_name="HP3_all_vocals.pth") for i, result in enumerate(results): wav_instrument, wav_vocals, sr, agg = result sf.write(f"{output_path}/instrument_{i}.wav", wav_instrument / 32768, sr) sf.write(f"{output_path}/vocals_{i}.wav", wav_vocals / 32768, sr) ``` but sadly I still get the same issue of the instrumental and vocals files being the same (original audio). Any help @Tps-F ?

bsakaguchi commented

2026-10-20 09:53:11 +00:00

Author

Owner

@Tps-F commented on GitHub (Apr 9, 2024):

I will check, It could be a bug.

@Tps-F commented on GitHub (Apr 9, 2024): I will check, It could be a bug.