Text to speech converter

Question

I am going to create a Text to speech converter, which can be executed in both Windows, and Linux based systems. Can someone please help me answer the following questions;

What language should i use; Java, C++ or any other?
The program i am going to create should be able to read large chunks of data.
There should be good sound quality.
What are the libraries available to achieve this?

Please share your experiences as well.

@Thorbjørn Ravn Andersen yes, and it should be free (open source or free to use once) — illep, Jul 01 '11 at 17:30
I think @Thorbjørn mean that you should look at a dead project or splinter off from a new one, as you'll be reinventing the wheel once more (but I don't discourage that at all. It's the best way to learn). — Blender, Jul 01 '11 at 17:32
This question was answered for Java in [this thread](http://stackoverflow.com/questions/143390/java-text-to-speech-engines-overview). There are many libraries. — toto2, Jul 01 '11 at 17:36
@Blender, no, I was thinking of simply just using an existing, living third party library. If it is a learning experience, it is another matter though. — Thorbjørn Ravn Andersen, Jul 02 '11 at 10:40

score 6 · Answer 1 · edited Jun 20 '20 at 09:12

Use the language you're most comfortable with. I learned the hard way when I tried to re-code my raytracer in C++, coming from a Python background. It took me weeks to get it to function...

Basically, you'll get frustrated by the nuances of a new language before you can get to coding something, so I suggest you stick with what you know.
That's independent of the language. Just makes sure to read the data in chunks (like 2048 kilobytes at a time). Reading a big file to RAM is not a great idea, as you can't be sure your end user has enough RAM (audio files can get BIG).
This is trivial. I can't help you program good sound quality. That's up to you.
There exist a few implementations of TTS for Linux, so that's all I can recommend. Look at Festival, as it is one of the ones you can actually understand.

javing · Answer 2 · 2011-07-01T18:41:53.897

Long time ago i had that same idea. I think it is not too complicated and can be done. This is how i would do it:

1- I would use Java, because it is platform independent.

2- If you are going to create an standalone application, it will be the local computer that will do the processing an not an application server, as in the case of web apps. Also i suppose that your program, will be used only by one user at a time. I think you will not have big issues with the amount data to process, if u implement some buffering mechanism.

3- The quality of the sound will depend in various factors such as: noise when recording,format, speed of the sound... Notice that many text to speech applications that exist on the web, the sound quality is ok, but you will never get the feeling that it is a real person, who is reading. (I recommend you a female voice, it sounds better).

4- Regarding to the libraries and another gadgets, this is what i think you might need:

-Java Standard Development Kit(To be able to work with java)

-A good IDE: There a many out there you can get for free. My favourite is eclipse but if you are a beginer with java, you might like to start with netbeans many people say it is more intuitive for learning.

-JLayer Mp3 library This will help you play the sounds in java.

-javax.swing Is a java library that is included in the development kit and allows you to create rich graphical user interfaces. Also if you are new to swing, i would recommend you a great swing IDE, called JFormDesigner, have a look at it, im sure it will speed your development alot(It is a great tool).

I hope my answers can give you some tips.

Update

I think it would be correct to mention also, that the only minimum requirement the users of your app will have in order to run the app is:

JRE(Java runtime enviroment)

@sfrj: But Java has a runtime requirement, and C++ can be compiled to run on almost any platform. — Blender, Jul 01 '11 at 18:28
@Blender The only requirement to run a java program that uses the gadgets i mentioned, is to have installed the java runtime enviroment in the machine, that is what makes java platform independent. I don't think it is a big effort for a user to install the JRE(Java Runtime Enviroment). Unfortunately windows does not have it installed by default, but there are some other systems that do have it, for example: Ubuntu. I never worked with C++, im sure it is a great language too, but i think java is also a really good choice for solving this problem. — javing, Jul 01 '11 at 18:37
@sfrj: Java is a great language for this, I agree, as many of the Speech to Text engines are written in it. I think Windows would be the only problematic OS, as it doesn't ship with the JRE. — Blender, Jul 01 '11 at 19:04

score 1 · Answer 3 · edited Mar 30 '12 at 09:30

1

This project of text to speech can be made using computer generated voice so no problem of storing a database for each word and its pronunciation. Just use the language in which u feel master to urself and apply the concept.....

edited Mar 30 '12 at 09:30

rve

5,897
3
40
64

answered Mar 30 '12 at 07:57

shakuntla yadav

11
1

score 0 · Answer 4 · answered Jul 01 '11 at 17:25

0

I do not know anything about text to speech except that it's not easy. That being said, I suggest you stick with the language you are most familiar with. I'm sure Java and C can do it.

answered Jul 01 '11 at 17:25

Joshua

4,270
10
42
62

Text to speech converter

4 Answers4