WhisperSpeech web UI

Web UI for WhisperSpeech (https://github.com/collabora/WhisperSpeech)

Info

Note

Versions 2.x, 3.x and 4.x allow voice generation via API.

Note

Version 4.x supports SillyTavern (1.15)

Test platforms:

Name	Info
CPU	AMD Ryzen 9 9950X3D
GPU	AMD Radeon 7900XTX
RAM	64GB DDR5 6600MHz
Motherboard	Gigabyte X870 AORUS ELITE WIFI7 (BIOS F8)
OS	Debian 13.4
Kernel	6.12.74+deb13+1-amd64
ROCm	7.2.3

Name	Info
CPU	IntelCore i5-12500H
GPU	NVIDIA GeForce RTX 4050
RAM	16GB DDR4 3200MHz
Motherboard	GIGABYTE G5 MF (BIOS FB10)
OS	CachyOS
Kernel	6.19.12-1-cachyos
NVIDIA Driver	595.58.03
CUDA	13.2

Instalation:

1. Install uv and ffmpeg.

2. Clone repository.

3. Mount the repository directory.

3. Create and activate venv using uv:

uv venv --python 3.13
source .venv/bin/activate

4. Run webui.py:

CPU (not recommended):

uv run --extra cpu webui.py

ROCm 7.2.3:

uv run --extra rocm webui.py

CUDA 13.2:

uv run --extra cuda webui.py

SillyTavern extension:

1. Copy whisperspeech-webui folder to SillyTavern/public/scripts/extensions/third-party

2. Run app with api support.

# This is example. If you are using SillyTavern locally, you can remove --listen parameter.
uv run --extra cpu webui.py --listen --api

3. Launch SillyTavern. In the extensions tab, expand WhisperSpeech web UI and set the IP and port.

GUI tanslation:

Languages
English
Polish

1. Extract messages.pot:

pybabel extract -F babel.cfg -o ./locale/messages.pot . 

2. Generate files:

New language:

pybabel init -i ./locale/messages.pot -d ./locale -l pl_PL
# Replace pl_PL by your language

Update current:

pybabel update -i ./locale/messages.pot -d ./locale -l pl_PL
# Replace pl_PL by your language

3. Compile:

pybabel compile -d ./locale