OPENVOICE API

An api engine for openvoice written in python

OpenVoice
https://github.com/myshell-ai/OpenVoice/tree/main

API Client
https://github.com/ground-creative/openvoice-api-client-python

OpenAI sdk example
https://github.com/ground-creative/openvoice-api-python/blob/main/examples/openai_lib_example.py

Installation

Docker

Follow instructions here to install with docker
https://github.com/ground-creative/openvoice-docker

Stand Alone Installation

You are going to need to include OpenVoice dependencies manually.

1. Clone the repository

git clone https://github.com/ground-creative/openvoice-api-python.git

2. Change environment variables in env.sample file and rename it to .env

Usage

#  install packages
pip install -r requirements.txt  # install dependencies

# Run the server
python3 app.py

Services

1. Generate speech

Method: POST

Endpoint: /{VERSION}/generate-audio

Params:

• model(required) the model to use
• input(required) the text to convert to speech
• speed(default: 1.0) the speed of the voice
• response_format(url|bytes|base64|stream)(default: url) the response format
• voice(default: raw) the voice to use

Extra params V1:

• style('default','whispering','shouting','excited','cheerful','terrified','angry','sad','friendly') a style for the voice

Extra params V2:

• accent(default: default language) an accent for the voice

2. Change voice of audio

Method: POST

Endpoint: /{VERSION}/change-voice

Params:

• model(required) the model to use
• audio_data(required) base64 encoded audio data
• voice(required) the voice to use
• response_format(url|bytes|base64|stream)(default: url) the response format

3. Retrieve a previously generated audio url

Method: GET

Endpoint: /audio-file/{FILENAME} Params:

• stream(true|false)(default: false)

Examples

Generate speech example

import requests, os

SERVER_PORT = os.getenv("SERVER_PORT", 5000)

output_file = 'outputs/generate_audio_url.wav'
if os.path.exists(output_file):
    os.remove(output_file)

version = 'v2'

# Define the URL of the generate_audio endpoint
url = f'http://localhost:{SERVER_PORT}/{version}/generate-audio'

# Define the parameters for the POST request
payload = {
    'model': 'en',
    'input': 'Hello, this is a test. I am here, there and everywhere. Let me know how you feel, perhaps this can be real.',
    'speed': 1.0,
    'voice': 'elon',
    'response_format': 'url',
    #'style': 'excited' # v1 only
    #'accent': 'en-au' #v2 only
}

try:
    # Send the POST request to generate the audio
    response = requests.post(url, json=payload, stream=False)
    if response.status_code == 200:   
        response_data = response.json()
        file_url = response_data['result']['data']['url']
        print(f'Generated url: {file_url}')
        response = requests.get(file_url, stream=False)
        if response.status_code == 200:
            with open(output_file, 'wb') as audio_file:
                audio_file.write(response.content)
            print(f'Audio file saved as {output_file}')
        else:
            print(f'Error getting file url: {response.status_code}')
            print(response.json())
    else:
        print(f'Error: {response.status_code}')
        print(response.json())

except requests.exceptions.RequestException as e:
    print(f'Request failed: {e}')

Generate speech with openai SDK example

import logging, os
from openai import OpenAI

SERVER_PORT = os.getenv("SERVER_PORT", 5000)
version = 'v2'
output_file = 'outputs/openai_lib_example.wav'
if os.path.exists(output_file):
    os.remove(output_file)

try:
    # Initialize OpenAI client
    client = OpenAI(
        base_url=f'http://localhost:{SERVER_PORT}/{version}',
        api_key='not-needed',  # Required but ignored by OpenAI
    )
    # Make request to OpenAI API
    with client.audio.speech.with_streaming_response.create(
        input='Hello, this is a test. I am here, there and everywhere',
        model='en',
        voice="elon",
        #extra_body={"accent": "en-au"} , # v2 only
        #extra_body={"style": "angry"} , # v1 only
        response_format="wav"
    ) as response:
        response.stream_to_file(output_file)

except Exception as e:
    logging.error(f"Unexpected error: {e}")

Look inside examples folder for more examples

Unit Testing

cd tests
python -m unittest __FILE__
# or
python -m unittest __FILE__.CLASS__
# or
python -m unittest __FILE__.CLASS__.FUNCTION__