What language to use for writing speech recognition software?

Question

I want to write basic Speech Recognition software that can convert speech to text. I wanted to know which language is most suitable for writing such software. Is Java suitable for this job?

edit: Thanks everyone for the answers. I want to create a tool for a college project. I do not want to write it from scratch. I just want to demonstrate the power of speech recognition. The tool should simply write everything the user says in a text editor such as notepad. This should not be too accurate. I just want to experiment and learn different speech recognition algorithms, because I find this field very interesting.

Thanks Deepak

Answer 1

Java may be appropriate for the interface, but speech recognition requires serious grunts. I would choose a compiled language similar to metal, such as C for a real recognition engine.

This is not something to be taken lightly, by the way. There is a lot of theory that you will need to study before you start. I myself, if possible, would get a license for one of the existing engines and focus on creating a decent product around it.

This is if you intend to build a product. If you just want to experiment, write yourself. It will be fun (to the point: -).

Answer 2

My students use Sphinx . It is written in Java (port from C ++, I believe). This may not be suitable for what you want (I think you will need to create your own dictionary), but it is worth checking out.

Answer 3

I agree with Pax that this is a potentially quite large project, and that the most practical solution is probably just a license for an existing engine.

If the volume of what you want to do simply distinguishes between several previously known possible statements, this is a much smaller project, but still significant.

But ... if you decide that you really really want to start developing your own, I see no reason not to use Java. The idea that "C is faster" is largely a myth (or based on outdated information).

Answer 4

Agree with almost everything Pax said, so I'm going to be the opposite and argue otherwise. Common wisdom is that speech recognition "requires serious grunts," and this may be because it is true.

But it can also happen that everyone believes that the way it was always done. Reasoning from the fact that the human brain does not make huge amounts of brute force data that are trying to recognize speech, I would suggest that there are smart feature extraction algorithms to make work much more efficient.

If so, and if you want to find such an algorithm, a higher-level language may be better suited to the task. Everything that you lose in efficiency, you will get in algorithmic expressiveness.

However, he is probably right.

Answer 5

I think java may be a good option, it all depends on how you get the input. Java has interesting libraries for sounds.

Language will not be a problem, because it will be a matter of pattern recognition. If java is the language you are most familiar with, I would use it.

Answer 6

Java turing completes so that it can handle every programming job. If you want to do something in Java, it is up to you.

Answer 7

We had moderate success with the Shynx framework written in Java, but the real hard work is to understand the algorithms and math involved in this area, and then fine-tune it for your specific needs.