Question

Most of you probably know the text-to-speech synthesizer of google translate, as you can access programmatically here btw:

http://translate.google.com/translate_tts?tl=en&q=text

My impression was it's sometimes using espeak, but in the major languages, the quality is much better than that. Anybody knows what Google is using, or what voices they are using? Clearly it's not the normal and also not the mbrola espeak voices.

Was it helpful?

Solution

I would look in the list of Google acquisitions
(Wikipedia, list of google acquisitions):

84 December 3, 2010 Phonetic Arts Speech synthesis  UK Google Voice, Google Translate [90]

OTHER TIPS

I have made a simple wrap with ruby. https://github.com/c2h2/tts

gem install tts
require 'tts'
'hello world!".to_file "en"

Try this one:

Speech Util

It´s free, but only for English.

I would be surprised if Google translate is using espeak. Firstly, the results are too good and lack many typical espeak flaws. Second, Google is well known to be using ideas like deep nets in their speech group (see the work by Geof Hinton and also http://research.google.com/pubs/SpeechProcessing.html).

Try pyttsx: https://github.com/parente/pyttsx

$ pip install pyttsx
$ python
>>> import pyttsx
>>> e = pyttsx.init()
>>> e.say('haha hahaha haha haha hahaha')
>>> e.runAndWait()

Use the pyttsx3 module for python3.

just use pip install pyttsx3 for installing

Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top