How Python implements text-to-speech

Get ready

I tested a python version of 2.7.10, and if your version is Python3.5, it's not a good place to be.

Using the speech API

Principle

Our idea is to use Microsoft's voice interface, so we're definitely going to call the relevant interface. So we need to install Pywin32 to help us complete this bottom-up interaction.

Sample code

123	importwin32com.clientspeaker =win32com.client.Dispatch("SAPI.SpVoice")speaker.Speak("Hello, it works!")

Small summary

Yes, it's so easy to invoke the interface to implement speech, but we have to talk about the drawbacks of this approach.

The Chinese support is not good enough, just this point, it is estimated that few in China to use it.

And the speed is not very good control

PYTTSX Way

Principle

PYTTSX is a great library of Python for text-to-speech. We can also use PYTTSX to read the RFC files or local files, and so on, most crucially, it is still good for Chinese support.

Sample code

12345678910111213141516

# coding:utf-8import sysreload(sys)sys.setdefaultencoding(‘utf8‘)# __author__ = ‘郭 璞‘# __date__ = ‘2016/8/6‘# __Desc__ = 文字转语音输出import pyttsxengine = pyttsx.init()engine.say(‘hello world‘)engine.say(‘你好，郭璞‘)engine.runAndWait()# 朗读一次engine.endLoop()

Small summary

With PYTTSX, we can use its powerful API to achieve our basic business needs. It's cool.

PYTTSX in-depth research

After doing the little experiment above, you will certainly feel how so not enjoyable?
Don't worry, let's go into Pyttsx's world together, and delve into how it works.

Voice engine Factory

Similar to "Factory mode" in design mode, PYTTSX is initialized to get the speech engine. When we call the init operation for the first time, we return a PYTTSX engine object, and when it is called again, if there is an Engine object instance, it will use the existing one, or else recreate it.

1	pyttsx.init([driverName : string, debug : bool]) → pyttsx.Engine

From a method declaration, the first parameter specifies the name of the voice driver, which is closely related to the underlying operating system. As follows:

1.drivename: called by the Pyttsx.driver module according to the operating system type, by default using the best driver that the current operating system can use

Sapi5-sapi5 on Windows

Nsss-nsspeechsynthesizer on Mac OS X

Espeak-espeak on every other platform

2.debug: This second parameter specifies whether you want to output in debug state, the recommended development stage is set to True

Engine interface

To use a library well, it is not possible to understand its API. Let's take a look at pyttsx. Engine. Engine API.

Method signature	Parameter list	return value	Simple explanation
Connect (topic:string, cb:callable)	Topic: The name of the event to describe; CB: callback function	→dict	Adds a callback notification on a given topic
Disconnect (token:dict)	Token: Return token for callback mismatch	Void	End Connection
Endloop ()	None	→none	In short, it's an end event loop.
GetProperty (name:string)	Name has these enumeration values "rate, vioce,vioces,volumn	→object	Gets the property value of the current engine instance
SetProperty (name:string)	Name has these enumeration values "rate, vioce,vioces,volumn	→object	Set property values for the current engine instance
Say (Text:unicode, name:string)	Text: The textual data to be read aloud; Name: Associated pronunciation people, generally not used	→none	Preset to read the text data, which is "Everything is ready, only the east wind" in the "Everything is ready"
Runandwait ()	None	→none	This method is the "East Wind". Returns when all events in the event queue are emptied
Startloop ([Usedriverloop:bool])	Usedriverloop: Whether drive cycle is enabled	→none	Turn on event queuing

Meta data tones

In Pyttsx.voice.Voice, handle the pronunciation of the synth.

Age

Pronunciation of the age of the person, default to None

Gender

Pronunciation by String type human gender: Male, female, or neutral. Default is None

Id

String acknowledgment information about the voice. Use Pyttsx.engine.Engine.setPropertyValue () to set the active pronunciation signature. This property is always defined.

Languages

The list of languages supported by the pronunciation, if not, is an empty list.

Name

Pronunciation person name, default to None.

More Tests

Read aloud text

12345

importpyttsxengine =pyttsx.init()engine.say(‘Sally sells seashells by the seashore.‘)engine.say(‘The quick brown fox jumped over the lazy dog.‘)engine.runAndWait()

Event Monitoring

12345678910

import pyttsxdef onStart(name): print ‘starting‘, namedef onWord(name, location, length): print ‘word‘, name, location, lengthdef onEnd(name, completed): print ‘finishing‘, name, completedengine = pyttsx.init()engine.say(‘The quick brown fox jumped over the lazy dog.‘)engine.runAndWait()

Interrupt pronunciation

12345678

import pyttsx def onword (name, location, length):   print ' word ' , name, location, length   if location > 10 :   engine.stop () engine = pyttsx.init () engine.say ( ) engine.runandwait ()

Change the voice of the Speaker

123456

engine = pyttsx.init () voices = engine.getproperty ( ' voices ' ) for voice in voices:   engine.setproperty ( , voice. id   engine.say ( Code class= "py string" > ' The quick brown fox jumped over the "the lazy dog. ' ) engine.runandwait ()

Speed control

12345

engine =pyttsx.init()rate =engine.getProperty(‘rate‘)engine.setProperty(‘rate‘, rate+50)engine.say(‘The quick brown fox jumped over the lazy dog.‘)engine.runAndWait()

Volume control

12345

engine =pyttsx.init()volume =engine.getProperty(‘volume‘)engine.setProperty(‘volume‘, volume-0.25)engine.say(‘The quick brown fox jumped over the lazy dog.‘)engine.runAndWait()

Perform an event-driven loop

1234567891011121314

engine = pyttsx.init()def onStart(name): print ‘starting‘, namedef onWord(name, location, length): print ‘word‘, name, location, lengthdef onEnd(name, completed): print ‘finishing‘, name, completed if name == ‘fox‘: engine.say(‘What a lazy dog!‘, ‘dog‘) elif name == ‘dog‘: engine.endLoop()engine = pyttsx.init()engine.say(‘The quick brown fox jumped over the lazy dog.‘, ‘fox‘)engine.startLoop()

Use an external drive loop

123456

engine =pyttsx.init()engine.say(‘The quick brown fox jumped over the lazy dog.‘, ‘fox‘)engine.startLoop(False)# engine.iterate() must be called inside externalLoop()externalLoop()engine.endLoop()

Summarize

The above is how Python implementation of text to speech all the content, read the above, is not the sense of Python to achieve text to speech or pretty simple? Well, let's try it. I hope this article will help you learn python.

Transferred from: http://www.jb51.net/article/90067.htm

How Python implements text-to-speech