[ISSUE for v2]: Stuttering and warning buzzers

[KSSSniper]

Voice Changer Version

vcclient_win_cuda_2.0.50-alpha

Operational System

Windows 10

GPU

RTX 4060

CUDA Version

32.0

Read carefully and check the options

• If you use win_cuda_torch_cuda edition, setup cuda? see here
• If you use win_cuda edition, setup cuda and cudnn? see here
• If you use mac edition, client is not launched automatically. Use chrome to open application.?
• I've tried to change the Chunk Size
• I've tried to set the Index to zero
• I've read the tutorial
• I've tried to extract to another folder (or re-extract) the .zip file

Does pre-installed model work?

YES

Model Type

RVC

Issue Description

Voice stutter and warning buzzers only in server mode.
It usually happen when I start the voice changing process, but it also randomly happen all of a sudden.
It occurs on different models, even pre-installed models.

The voice quality seems to fluctuate every time I start the voice changing process. (Sometimes higher pitched, sometime lower pitched, without tempering with the actual pitch bar of course)

CPU is at 14% and Memory at around 62%, GPU at 21%. No sudden spikes can be observed during sutter or warning noise.

Application Screenshot

No response

Logs on console

Log seems to be filled with similar Debug lines
2024-11-09 14:42:39,293 - vcclient - local_voice_changer_ - INFO - audio_output_callback trancate queue: 140 - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 252
2024-11-09 14:42:39,419 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:39,541 - vcclient - local_voice_changer_ - INFO - audio_output_callback trancate queue: 156 - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 252
2024-11-09 14:42:39,649 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:39,757 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:39,927 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:40,036 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:46,483 - vcclient - local_voice_changer_ - INFO - audio_output_callback trancate queue: 86 - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 252
2024-11-09 14:42:46,596 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:46,704 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:46,874 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:46,982 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:47,106 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:47,307 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:48,164 - vcclient - local_voice_changer_ - INFO - audio_output_callback trancate queue: 86 - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 252
2024-11-09 14:42:48,265 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:48,374 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:48,560 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:48,668 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:48,779 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:49,736 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:50,507 - vcclient - local_voice_changer_ - INFO - Start Local Voice Changer(1way) counter:100, sr:[44100.0, 44100.0], ch:[2, 2], depth:[float32, float32] - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 188
2024-11-09 14:42:50,537 - vcclient - local_voice_changer_ - INFO - audio_output_callback trancate queue: 94 - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 252
2024-11-09 14:42:50,649 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:50,758 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:50,928 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:51,037 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:51,145 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:52,132 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:52,945 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263
2024-11-09 14:42:54,534 - vcclient - local_voice_changer_ - DEBUG - audio output callback ex: empty queue : - vcclient_dev\voice_changer\voice_change_manager\local_voice_changer_interface.py - 263

[Kuuko-fokkusugaru]

First try updating to the latest version.

[KSSSniper]

The problem seems to persit after updating to the latest version.
Sometimes the quality of the voice is excellent and sometimes is poor, the quality of the voice seems to be different everytime I hit start. It's like a russian roulette of quality.
The latency also seems to increase upon updating (With all chunk/extra settings staying the same)

The good quality seems only achievable via server mode.

[Kuuko-fokkusugaru]

Does the quality is bad through monitor or in other apps using output?
Are you doing another tasks while using RVC like playing a game?
If you could provide a screenshot of the software and a little clip to listen how exactly sounds wrong I may be able to identify the issue better.

[Kuuko-fokkusugaru]

Please take the screenshot while speaking and not in idle.

[KSSSniper]

Thanks for the reply!
I will try to reenact the good quality and the bad quality and voice and record it tommorow.

I was using Voicemeeter Banana to monitor the voice, might try using the monitor on the client to see if there's any different.
But when the quality is bad, it's kinda constant through monitor and output (discord/Vrchat), vice versa when it's good.

I usually start the RVC before games, but I didn't notice and change of quality during gaming sessions, it quality remain the same with or without the game. Like, the quality won't drop to low quality or improve mid game, but when I start/stop the RVC process, or I temper with the chunks/extra. (Minor changes of course, like 0.8-->0.9 sec, which supposingly shouldn't cause it to change drastically)

Oh, my headset is using a 3.5 jack but my PC's jack doesn't work, so I am using an usb adapter. (If this could be the problem)

[Kuuko-fokkusugaru]

Voice meter, vb cable, or any other software of that developer gives quality issues with some software and games. Try instead the free Virtual Audio Cable Lite (VAC). That one works wonders and don't need to setup anything beyond just installing

[KSSSniper]

Here is the screenshot of the Client and two clips attatched.

(I was using Crepe when I took the screenshot but usually I hop around Rmvpe and Crepe to see which yield better results)

(The audio clips usually only last around 15 seconds, the conversion makes it extra long w)
There's warning beeps at the beginning of the audio, they sometime come randomly with stuttering and cutting off the output.
(Mid-sentence ----> Audio cutoff ----> BEEP BEEP -----> Resume Normally)
https://github.com/user-attachments/assets/b8893b8-b00f-43de-89cc-7344a775ec88

Here's a clip showing how the sound changes. ( I stop the process and restarted it, but the voice just changes)
https://github.com/user-attachments/assets/6540868-8965-4663-b75d-65b7ecf2b733
(Pardon for my bad English)

Cheers.

[Kuuko-fokkusugaru]

You are still using voice meter which I already mentioned that it gives quality issues. I even could listen both of the voices, original and converted, which proves that the setup isn't right as that shouldn't happen. Clips should always be recorded from the output to get a clear example of what's exactly coming out. The beep alert sound it's something else though. I am not sure what could be but it sounds intentional as if you are using the demo of a paid software. It's voice meter free? Aren't you using a paid demo version by mistake instead of a free version? Still, you will keep getting quality issues with it. Use VAC Lite or you won't be able to use RVC properly in different games or Discord. If you want to keep using voice meter then you may need to contact their support about the quality issues because I can hear that it sounds fine except for the pitch change after restart. That could be a bug. You may need to test if it happens on prior versions because it could be a new one. You could also give a try to v1.5.3.18a and see if it works better.

[KSSSniper]

I was using Canatabile as VST host so I need some time to set things up with VAC cables but will eventually switch to them.
...and YEAH, thank god I did the recording, I didn't set some of the things clearly.

Regarding the beeping, I doubt it was coming from Voicemeeter as it was only audible in server mode. The beeping was also noticed in the build-in monitoring as well. Gonna try looking into the hardwares and VST plugins.

I will give MMVC version a try as well, bit switching to VST host will be first step I guess.
Cheers!

[Kuuko-fokkusugaru]

The beep sound can't come from RVC directly so it may be a driver issue of another kind. The reason is that RVC would convert the beep sound into voice if it were to be fed through the input and it clearly sounds like a beep alert. If it comes through monitor then the beep is happening at output level, meaning your speakers/headset. Don't use sever mode as that's meant only when using RVC on a secondary computer.