Compiling Sphinx3 for Win32. VS .sln file missing

[videohead]

Hi -

I'm trying to compile Sphinx3 for Win 32. Unfortunately the source files
do not include a .sln file. Please advise.

Regards,

• Olumide

[Nickolay V. Shmyrev]

Use pocketsphinx

[videohead]

Thanks for your reply. I've got several questions:

1. Does PocketSphinx perform phoneme recognition?

2. Are there any tutorials for PocketSphinx? The only tutorials that I've found are for Shpinx3.

3. I notice that SphinxTrain and SphinxBase appear to contain the same binaries as PocketSphinx. Are SphinxTrain and SphinxBase supersets of PocketSphinx?

4. I am looking to adapt the acoustic models provided to British English. Which of the three projects should I be using?

5. Can adaptation be performed simultaneously using data from more than 1 speaker, or even incrementally? i.e. can an adapted model be readapted at a latter time?

[Nickolay V. Shmyrev]

Does PocketSphinx perform phoneme recognition?

No

Are there any tutorials for PocketSphinx? The only tutorials that I've found are for Shpinx3.

http://cmusphinx.sourceforge.net/wiki/tutorial

I notice that SphinxTrain and SphinxBase appear to contain the same binaries as PocketSphinx. Are SphinxTrain and SphinxBase supersets of PocketSphinx?

You can learn about package hierarchy from the tutorial

http://cmusphinx.sourceforge.net/wiki/tutorialoverview

I am looking to adapt the acoustic models provided to British English. Which of the three projects should I be using?

You need all three

Can adaptation be performed simultaneously using data from more than 1 speaker, or even incrementally? i.e. can an adapted model be readapted at a latter time?

Yes

[videohead]

Thanks. To recap, bearing in mind that I am interested in phoneme recognition, should I stick to SphinxTrain, SphinxBase and Shinx3 (I recall reading somewhere that Sphinx4 does not perform phoneme recognition), or has all the functionality of Shinx3 been moved to SphinxTrain, SphinxBase?

[videohead]

As a follow up, the Phoneme Recognition page of the wiki http://cmusphinx.sourceforge.net/wiki/phonemerecognition references sphinx3_decoder.exe. No such similarly-named binary (decoder.exe) is contained in SphinxTrain, SphinxBase or PocketSphinx .

[Nickolay V. Shmyrev]

I am interested in phoneme recognition

Phoneme recognition is almost never accurate, for that reason most likely you do not need it.

has all the functionality of Shinx3 been moved to SphinxTrain, SphinxBase

No

[videohead]

I am aware that the lack of text constraints makes phoneme recognition inaccurate. Unfortunately my application requires phoneme transcriptions. The emphasis is not merely on what the intends to say but on what is actually said, for example elided or mispronounced words etc. I guess I should give Sphinx3 a try and see just how accurate it is and go on from there. Can you provide the *.sln file for Sphinx3? It is missing from the download.

Thanks

[Nickolay V. Shmyrev]

sln file is available in subversion, for more details see:

https://sourceforge.net/p/cmusphinx/code/

[videohead]

Thanks.