Text to Speech in Python for NLP: Simple Guide & Example

You can do text to speech in Python using the pyttsx3 library, which converts text into spoken audio offline. Install it with pip install pyttsx3, then use its simple API to input text and play or save the speech output.

Syntax

The basic syntax to use pyttsx3 involves initializing the engine, setting properties like speech rate or voice, and then calling say() to queue text for speaking. Finally, runAndWait() processes the speech.

• engine = pyttsx3.init(): Creates the speech engine.
• engine.say(text): Adds text to speak.
• engine.runAndWait(): Runs the speech queue and waits until done.
• Optional: engine.setProperty() to adjust voice or speed.

python

import pyttsx3

engine = pyttsx3.init()  # Initialize the TTS engine
engine.say('Hello, this is a test.')  # Queue text to speak
engine.runAndWait()  # Play the speech

Output

The computer speaks: "Hello, this is a test."

Example

This example shows how to convert a simple text string to speech, change the speech rate, and select a different voice if available.

python

import pyttsx3

engine = pyttsx3.init()

# Set speech rate (default is usually 200 words per minute)
rate = engine.getProperty('rate')
engine.setProperty('rate', rate - 50)  # slower speech

# List available voices and select one
voices = engine.getProperty('voices')
if len(voices) > 1:
    engine.setProperty('voice', voices[1].id)  # change to second voice if exists

text = 'Welcome to text to speech conversion using Python.'
engine.say(text)
engine.runAndWait()

Output

The computer speaks: "Welcome to text to speech conversion using Python."

Common Pitfalls

Common mistakes include not calling runAndWait() after say(), which means the speech won't play. Another issue is using voices that may not exist on your system, causing errors. Also, forgetting to install pyttsx3 or missing dependencies can cause failures.

Always test voice availability and handle exceptions for better reliability.

python

import pyttsx3

engine = pyttsx3.init()

# Wrong: Missing runAndWait(), speech will not play
engine.say('This will not be heard')

# Correct:
engine.say('This will be heard')
engine.runAndWait()

Output

The computer speaks only: "This will be heard"

Quick Reference

Function	Description
pyttsx3.init()	Initialize the text-to-speech engine
engine.say(text)	Queue text to be spoken
engine.runAndWait()	Process and play the queued speech
engine.setProperty('rate', value)	Set speech speed (words per minute)
engine.setProperty('voice', voice_id)	Set the voice by ID
engine.getProperty('voices')	Get list of available voices

Key Takeaways

Use the pyttsx3 library for offline text to speech in Python.

Always call engine.runAndWait() after engine.say() to hear speech.

Adjust speech rate and voice properties to customize output.

Check available voices on your system before setting voice.

Install pyttsx3 with pip and ensure dependencies are met.