streamlit-TTS

Streamlit component providing tools to convert text to speech and autoplay audio directly in the browser without any visible UI elements.

Features

• UI-less Audio Autoplay: Plays audio invisibly without disrupting your app's layout
• Cross-Browser Compatibility: Robust autoplay with fallback strategies for Chrome, Safari, Firefox, and mobile browsers
• Dual TTS Engines: Support for both Google TTS (gTTS) and OpenAI TTS
• Smart Text Cleanup: Automatically removes markdown, LaTeX, code snippets, and URLs before synthesis
• Type-Safe: Full type hints for better IDE support and code quality

Installation

pip install streamlit-TTS

API Reference

Five main functions are provided:

1. auto_play(audio, wait=True, lag=0.25, key=None)

Plays audio directly in the browser without showing any UI element.

Parameters:

• audio (Optional[Dict]): Audio dictionary with keys 'bytes', 'sample_rate', and 'sample_width', or None
• wait (bool): If True, blocks execution until audio finishes playing (default: True)
• lag (float): Additional wait time in seconds to account for network latency (default: 0.25)
• key (Optional[str]): Unique key for the component

Audio Dictionary Format:

audio = {
    'bytes': bytes,              # WAV audio bytes
    'sample_rate': int,          # Sample rate (e.g., 44100)
    'sample_width': int,         # Sample width in bytes (e.g., 2)
}

Example:

from streamlit_TTS import auto_play
auto_play(audio_dict, wait=True)

2. text_to_audio(text, language='en', cleanup_hook=None)

Converts text to audio using Google Text-to-Speech (gTTS).

Parameters:

• text (str): The text to convert to speech
• language (str): Language code (e.g., 'en', 'fr', 'es') (default: 'en')
• cleanup_hook (Optional[Callable]): Custom text cleaning function (default: built-in cleanup)

Returns: Audio dictionary or None if text is empty

Example:

from streamlit_TTS import text_to_audio
audio = text_to_audio("Hello world!", language='en')

3. text_to_speech(text, language='en', wait=True, lag=0.25, key=None)

Convenience function combining text_to_audio() and auto_play() using Google TTS.

Parameters:

• text (str): The text to convert to speech
• language (str): Language code (default: 'en')
• wait (bool): If True, blocks until audio finishes (default: True)
• lag (float): Additional wait time in seconds (default: 0.25)
• key (Optional[str]): Unique key for the component

Example:

from streamlit_TTS import text_to_speech
text_to_speech("Welcome to my app!", language='en')

4. openai_text_to_audio(text, openai_api_key=None, language=None, cleanup_hook=None, voice='shimmer', model='gpt-4o-mini-tts')

Converts text to audio using OpenAI's Text-to-Speech API.

Parameters:

• text (str): The text to convert to speech
• openai_api_key (Optional[str]): OpenAI API key (uses OPENAI_API_KEY env var if not provided)
• language (Optional[str]): Language code (currently unused by OpenAI API)
• cleanup_hook (Optional[Callable]): Custom text cleaning function
• voice (str): Voice name - 'alloy', 'echo', 'fable', 'onyx', 'nova', or 'shimmer' (default: 'shimmer')
• model (str): Model to use - 'gpt-4o-mini-tts', 'gpt-4o-audio-preview', 'tts-1', or 'tts-1-hd' (default: 'gpt-4o-mini-tts')

Returns: Audio dictionary or None if text is empty

Example:

from streamlit_TTS import openai_text_to_audio
audio = openai_text_to_audio("Hello!", voice='nova', model='gpt-4o-mini-tts')

5. openai_text_to_speech(text, language=None, wait=True, lag=0.25, key=None, voice='shimmer', model='gpt-4o-mini-tts')

Convenience function combining openai_text_to_audio() and auto_play().

Parameters:

• text (str): The text to convert to speech
• language (Optional[str]): Language code (currently unused)
• wait (bool): If True, blocks until audio finishes (default: True)
• lag (float): Additional wait time in seconds (default: 0.25)
• key (Optional[str]): Unique key for the component
• voice (str): Voice name (default: 'shimmer')
• model (str): Model to use (default: 'gpt-4o-mini-tts')

Example:

from streamlit_TTS import openai_text_to_speech
openai_text_to_speech("Welcome!", voice='nova')

Examples

Basic Usage (Google TTS)

import streamlit as st
from streamlit_TTS import text_to_speech

# Simple text-to-speech
text_to_speech("Hello! Welcome to my Streamlit app!", language='en')

# With user input
text = st.text_input("Enter text to speak:")
if st.button("Speak") and text:
    text_to_speech(text, language='en')

Using OpenAI TTS

import streamlit as st
from streamlit_TTS import openai_text_to_speech
import os

# Set your OpenAI API key
os.environ['OPENAI_API_KEY'] = 'your-api-key-here'

# Use OpenAI TTS with different voices
voice = st.selectbox("Choose voice", ['alloy', 'echo', 'fable', 'onyx', 'nova', 'shimmer'])
text = st.text_input("Enter text:")

if st.button("Speak") and text:
    openai_text_to_speech(text, voice=voice, model='gpt-4o-mini-tts')

Advanced Usage with Custom Cleanup

import streamlit as st
from streamlit_TTS import text_to_audio, auto_play

# Custom text cleanup function
def my_cleanup(text):
    # Remove specific patterns
    text = text.replace("[SPEAKER]", "")
    return text.strip()

# Generate audio with custom cleanup
audio = text_to_audio(
    "Some text with [SPEAKER] tags",
    language='en',
    cleanup_hook=my_cleanup
)

# Play the audio
if audio:
    auto_play(audio, wait=True, lag=0.5)

Multi-Language Example

import streamlit as st
from streamlit_TTS import text_to_speech
from gtts.lang import tts_langs

# Get available languages
langs = tts_langs()

# Welcome message
text_to_speech(
    "Choose a language, type some text, and click 'Speak it out!'",
    language='en'
)

# Language selection
lang = st.selectbox("Choose a language", options=list(langs.keys()))
text = st.text_input("Enter text to speak:")

if st.button("Speak it out!") and lang and text:
    text_to_speech(text=text, language=lang)

Browser Compatibility

This component includes robust autoplay with fallback strategies for maximum browser compatibility:

• ✅ Chrome (Desktop & Mobile)
• ✅ Safari (Desktop & Mobile)
• ✅ Firefox
• ✅ Edge
• ✅ Opera

The component automatically handles browser autoplay restrictions with multiple fallback strategies.

Requirements

• Python >= 3.7
• streamlit >= 0.63
• gtts
• pydub
• openai (optional, for OpenAI TTS features)

License

MIT