Using pyttsx3 for Text-to-Speech in Python

The author explores how to convert text to speech using the offline Python library pyttsx3.

Installation is done via pip: pip3 install pyttsx3 Basic usage involves initializing the engine, queuing text with say(), and invoking runAndWait() to produce audio:

import pyttsx3
engine = pyttsx3.init()
engine.say('Hello world')
engine.runAndWait()

The script demonstrates how to inspect and adjust voice properties such as rate, volume, and available voices:

rate = engine.getProperty('rate')
engine.setProperty('rate', 200)
volume = engine.getProperty('volume')
engine.setProperty('volume', 1.0)
voices = engine.getProperty('voices')
for v in voices:
    print(v.id, v.name)

For Chinese speech, no explicit language setting is required; the engine works with Mandarin text directly:

engine.say('语音合成开始')
engine.say('我会说中文了，开森，开森')
engine.runAndWait()

A more elaborate example shows conditional speech based on lists of known and unknown names, with different intervals for each case:

import pyttsx3
engine = pyttsx3.init()
def VOICE_TEXT(names):
    unknown_name = []
    known_name = []
    for name in names:
        if name == 'Unknown':
            unknown_name.append(name)
        else:
            known_name.append(name)
    num_known = len(known_name)
    num_unknown = len(unknown_name)
    if len(names) == 0:
        return
    elif num_unknown == 0:
        text = '欢迎'
        for i in range(num_known):
            if i + 1 == num_known and i != 0:
                text += '和{}'.format(known_name[i])
            else:
                text += known_name[i]
        engine.say(text)
        engine.runAndWait()
    elif num_known == 0:
        text = '警告，警告，有陌生人未识别，有陌生人未识别'
        engine.say(text)
        engine.runAndWait()
    else:
        text = '欢迎'
        for i in range(num_known):
            if i + 1 == num_known and i != 0:
                text += '和{}'.format(known_name[i])
            else:
                text += known_name[i]
        engine.say(text)
        engine.say('请注意， 您旁边有未识别人员，您将不允许进入，请询问值班人员，谢谢合作')
        engine.runAndWait()
if __name__ == '__main__':
    VOICE_TEXT(['张三丰', '张无忌'])
    VOICE_TEXT(['Unknown', '张三丰'])
    VOICE_TEXT(['Unknown'])

An advanced version adds global counters to limit how often each speech type is spoken, resetting after reaching thresholds:

import pyttsx3
KUNOWN = 0
UNKNOWN = 0
UNKNOWN_KNOWN = 0
engine = pyttsx3.init()
def VOICE_TEXT(names):
    unknown_name = []
    known_name = []
    for name in names:
        if name == 'Unknown':
            unknown_name.append(name)
        else:
            known_name.append(name)
    num_known = len(known_name)
    num_unknown = len(unknown_name)
    if len(names) == 0:
        return
    elif num_unknown == 0:
        global KUNOWN
        KUNOWN += 1
        if KUNOWN == 15:
            KUNOWN = 1
            text = '欢迎'
            for i in range(num_known):
                if i + 1 == num_known and i != 0:
                    text += '和{}'.format(known_name[i])
                else:
                    text += known_name[i]
            engine.say(text)
            engine.runAndWait()
    elif num_known == 0:
        global UNKNOWN
        UNKNOWN += 1
        if UNKNOWN == 10:
            UNKNOWN = 1
            text = '警告，警告，有外来人员未识别，有外来人员未识别'
            engine.say(text)
            engine.runAndWait()
    else:
        global UNKNOWN_KNOWN
        UNKNOWN_KNOWN += 1
        if UNKNOWN_KNOWN == 25:
            UNKNOWN_KNOWN = 2
            text = '欢迎'
            for i in range(num_known):
                if i + 1 == num_known and i != 0:
                    text += '和{}'.format(known_name[i])
                else:
                    text += known_name[i]
            engine.say(text)
            engine.say('请注意， 您旁边有外来人员，您将不允许进入，请询问值班人员，谢谢合作')
            engine.runAndWait()
if __name__ == '__main__':
    VOICE_TEXT(['张三丰', '张无忌'])
    VOICE_TEXT(['Unknown', '张三丰'])
    VOICE_TEXT(['Unknown'])

The tutorial concludes by suggesting that the speech function can be called from a timed loop in a larger program, using the counter variables to control playback frequency.