12

Can I add the dictionary of my own voice for text to speech application? If it possible which development language would be best to develop such a this kind of application There are several online tool for text to speech but they have there own voices. I have to develop same like that application.please suggest me how I can go ahead with my concept .

user3942918
  • 25,539
  • 11
  • 55
  • 67
Amit Golhar
  • 799
  • 4
  • 8
  • 21
  • Are you using native `window.speechSynthesis` for the text-to-speech application? – guest271314 Jul 15 '17 at 19:01
  • No I'm thinking to develop new application for that, but if it possible using windows speech synthesis then I can go through that – Amit Golhar Jul 15 '17 at 19:03
  • I have to also save that speech for further use – Amit Golhar Jul 15 '17 at 19:04
  • This Question/Answer should address that topic of recording and saving the generated audio result of `window.speechSynthesis.speak()` call [How to capture generated audio from window.speechSynthesis.speak() call?](https://stackoverflow.com/questions/45003548/how-to-capture-generated-audio-from-window-speechsynthesis-speak-call); using `meSpeak.js` and other approaches [How to create or convert text to audio at chromium browser?](https://stackoverflow.com/questions/44346410/how-to-create-or-convert-text-to-audio-at-chromium-browser) – guest271314 Jul 15 '17 at 19:08
  • Technically, yes, you can record your own voice and create a dictionary of key, value pairs reflecting the word and the audio file of the voice for that word. That would encompass essentially using the approaches at https://stackoverflow.com/questions/44346410/how-to-create-or-convert-text-to-audio-at-chromium-browser – guest271314 Jul 15 '17 at 19:16
  • Is it possible to add my own dictionary so that I can get the output with my voice . The input text will be anything such as it me will be one document – Amit Golhar Jul 15 '17 at 19:17
  • What would be the process to add the dictionary of my voice? Can you pls give some small basic idea – Amit Golhar Jul 15 '17 at 19:18
  • Yes. Record your voice for each word that you want the application to render as audio. Create an object of key, value pairs reflecting the word (property) and value (`data URL` or path to audio resource). – guest271314 Jul 15 '17 at 19:19
  • Need to record only alphabet A to Z or whole dictionary which has N number of words – Amit Golhar Jul 15 '17 at 19:21
  • _"Need to record only alphabet A to Z or whole dictionary which has N number of words "_ Yes, that is possible. In fact, that is one of the motivations for composing [SpeechSynthesisRecorder](https://github.com/guest271314/SpeechSynthesisRecorder), see [Re: MediaStream, ArrayBuffer, Blob audio result from speak() for recording?](https://lists.w3.org/Archives/Public/public-speech-api/2017Jul/0004.html) – guest271314 Jul 15 '17 at 19:23

2 Answers2

19

You can build your voice with open source software like Festival or Openmary. You will need a carefully prepared recording of your voice for about 1 hour

There are also commercial services which allows you to build a custom voice, for example Cereproc

Update 2019: These days you can use modern neural network toolkits see Nvidia Tactoron2 and Realtime Voice Cloning

Nikolay Shmyrev
  • 24,897
  • 5
  • 43
  • 87
-5

You can record your voice for each word that you want the application to render as audio. Create an object of key, value pairs reflecting the word (property) and value (data URL or path to audio resource).

const voices = { 
                        // or path to static file
                 hello: "data:audio/ogg,/* base64 string of your voice */" 
               }

_speak("hello") // output audio, that is, your voice
guest271314
  • 1
  • 15
  • 104
  • 177