[ISSUE]: Very High Res

[Kkenzzio]

Voice Changer Version

MMVCServerSIO_win_onnxdirectML-cuda_v.1.5.3.15.zip

Operational System

Windows 11

GPU

NVIDIA Gefore RTX 3080 Ti

Read carefully and check the options

• I've tried to Clear Settings
• Sample/Default Models are working
• I've tried to change the Chunk Size
• GUI was successfully launched
• I've read the tutorial
• I've tried to extract to another folder (or re-extract) the .zip file

Model Type

RVC

Issue Description

So basically,the issue here is when i sit for example at 96 Chunk Size,my res is normal (around 100 to 200 ms) but if i switch to anything lower like 80 or 64 (and i'm pretty sure my GPU can handle that type of load) it flies up to 10k+ ms res, and i tried many things but so far nothing helped.

Thanks in advance

Application Screenshot

Logs on console

D:\VoiceAI\MMVCServerSIO>MMVCServerSIO.exe -p 18888 --https false --content_vec_500 pretrain/checkpoint_best_legacy_500.pt --content_vec_500_onnx pretrain/content_vec_500.onnx --content_vec_500_onnx_on true --hubert_base pretrain/hubert_base.pt --hubert_base_jp pretrain/rinna_hubert_base_jp.pt --hubert_soft pretrain/hubert/hubert-soft-0d54a1f4.pt --nsf_hifigan pretrain/nsf_hifigan/model --crepe_onnx_full pretrain/crepe_onnx_full.onnx --crepe_onnx_tiny pretrain/crepe_onnx_tiny.onnx --rmvpe pretrain/rmvpe.pt --model_dir model_dir --samples samples.json
Booting PHASE :main
PYTHON:3.10.11 (tags/v3.10.11:7d4cc5a, Apr 5 2023, 00:38:17) [MSC v.1929 64 bit (AMD64)]
Activating the Voice Changer.
[Voice Changer] download sample catalog. samples_0003_t2.json
[Voice Changer] download sample catalog. samples_0003_o2.json
[Voice Changer] download sample catalog. samples_0003_d2.json
[Voice Changer] model_dir is already exists. skip download samples.
Internal_Port:18888
protocol: HTTP
-- ---- --
Please open the following URL in your browser.
http://:/
In many cases, it will launch when you access any of the following URLs.
http://127.0.0.1:18888/

gin_channels: 256 self.spk_embed_dim: 109
[VCClient] Access http://127.0.0.1:18888/
[VCClient] wait web server...0 http://127.0.0.1:18888/
[Voice Changer] generate new embedder. (no embedder)
[VCClient] wait web server...10 http://127.0.0.1:18888/
[Voice Changer] Loading index...
Try loading... model_dir\6\added_IVF595_Flat_nprobe_1_OMEN V3_v2.index
[VCClient] wait web server...20 http://127.0.0.1:18888/
[VCClient] wait web server... done 200
[2025-02-19 20:47:06] connet sid : 5i0D7TspLg1n0Fp5AAAB
[2025-02-19 20:47:07] connet sid : GlB4ZlofQi4_LNlKAAAD
[Voice Changer] update configuration: serverReadChunkSize 80
[Voice Changer] update configuration: serverReadChunkSize 48
[Voice Changer] update configuration: serverReadChunkSize 80
[Voice Changer] update configuration: serverReadChunkSize 64
[2025-02-19 20:56:43] connet sid : 6fR64vPJGMND8XKXAAAF

[Kuuko-fokkusugaru]

Which other tasks are you doing while using the software?

[Kuuko-fokkusugaru]

I can see many issues here.
First, update to the latest version. Even if you want to keep using v1 over v2, yours is outdated. The latest v1 is v1.5.3.18a.
Then download the right version for your system. You are using the direct ml version which is mostly for cpu usage and AMD cards. Download the CUDA version so you can make proper usage of your gpu.
That said, gpu usage will always increase exponentially when lowering the chunk size. Besides gpu usage, quality will also lower as there is less time to process the audio. I recommend a chunk size of 128 which should be around 0.37 seconds. 0.5 seconds is a good value too as it gives even better results. You can't get rid of latency if you don't want to sound bad.

[Kkenzzio]

I use it while my pc is idle, and there is nothing else running on pc expect few background processes.

Alright,I'll try those out,but mind telling me how to use CUDA if it's fine?

[Kuuko-fokkusugaru]

You just simply download the CUDA version instead is the direct ml version. You are simply using the wrong one which won't take full advantage of your GPU.

[Kkenzzio]

Alright,thanks.