pyVideoTrans

Appearance

Kokoro TTS Voice Channel ​

What is it?

Kokoro TTS is a lightweight voice synthesis project supporting 8 languages: Chinese, English, Japanese, French, Italian, Portuguese, Spanish, and Hindi. It provides a web interface and API endpoints.

Advantages:

  • Supports 8 languages
  • Web interface for previewing audio
  • Compatible with OpenAI API format
  • Lightweight and fast to start

Kokoro TTS Web interface

Project Information ​

Installation ​

Windows Users ​

  1. Download the package: https://github.com/jianchang512/kokoro-uiapi/releases/v0.1
  2. Extract and double-click start.bat to launch
  3. For GPU acceleration, make sure you have an NVIDIA GPU with CUDA 12 installed

Linux/macOS Users ​

Prerequisites:

  • Python 3.8+ (3.10-3.11 recommended)
  • ffmpeg installed:
    • Linux: apt install ffmpeg or yum install ffmpeg
    • macOS: brew install ffmpeg

Installation Steps:

bash
# 1. Clone the repository
git clone https://github.com/jianchang512/kokoro-uiapi

# 2. Create and activate a virtual environment
cd kokoro-uiapi
python3 -m venv venv
source venv/bin/activate

# 3. Install dependencies
pip3 install -r requirements.txt

# 4. Start
python3 app.py

Using It in pyVideoTrans ​

Step 1: Start Kokoro TTS ​

  • Windows package: double-click start.bat
  • Source install: run python3 app.py

Step 2: Configure pyVideoTrans ​

  1. Upgrade pyVideoTrans to v3.48+ or later
  2. Go to Menu → TTS Settings → Kokoro TTS
  3. In the HTTP address field, enter: http://127.0.0.1:5066
  4. Click the test button to verify the connection
  5. Save the settings

Enter Kokoro TTS address

Step 3: Start Dubbing ​

  1. Return to the main interface
  2. Select Kokoro TTS from the Voice Channel dropdown
  3. Choose the target language and voice
  4. Click the dubbing button to start

OpenAI API Compatibility ​

Kokoro TTS's API is compatible with the OpenAI TTS format, so you can use the OpenAI SDK to call it.

API Details ​

  • Endpoint: http://127.0.0.1:5066/v1/audio/speech
  • Method: POST
  • Content-Type: application/json

Request Parameters ​

json
{
    "input": "Text to synthesize",
    "voice": "Voice name",
    "speed": 1.0
}

OpenAI SDK Example ​

python
from openai import OpenAI

client = OpenAI(
    api_key='123456',
    base_url='http://127.0.0.1:5066/v1'
)

try:
    response = client.audio.speech.create(
        model='tts-1',
        input='Hello, my dear friends',
        voice='zf_xiaobei',
        response_format='mp3',
        speed=1.0
    )
    with open('./test_openai.mp3', 'wb') as f:
        f.write(response.content)
    print("MP3 file saved successfully")
except Exception as e:
    print(f"An error occurred: {e}")

Voice List ​

English Voices ​

af_alloy, af_aoede, af_bella, af_jessica, af_kore, af_nicole, af_nova,
af_river, af_sarah, af_sky, am_adam, am_echo, am_eric, am_fenrir,
am_liam, am_michael, am_onyx, am_puck, am_santa, bf_alice, bf_emma,
bf_isabella, bf_lily, bm_daniel, bm_fable, bm_george, bm_lewis

Chinese Voices ​

zf_xiaobei, zf_xiaoni, zf_xiaoxiao, zf_xiaoyi,
zm_yunjian, zm_yunxi, zm_yunxia, zm_yunyang

Japanese Voices ​

jf_alpha, jf_gongitsune, jf_nezumi, jf_tebukuro, jm_kumo

Other Language Voices ​

  • French: ff_siwis
  • Italian: if_sara, im_nicola
  • Hindi: hf_alpha, hf_beta, hm_omega, hm_psi
  • Spanish: ef_dora, em_alex, em_santa
  • Portuguese: pf_dora, pm_alex, pm_santa

Proxy/VPN ​

The source deployment method requires downloading voice pt files from Hugging Face, so you may need to set up a proxy or VPN beforehand.

Alternatively, you can download the models in advance and extract them to the app.py directory:

Credits ​