I need help!

[Khurram Shehzad]

Hi,
I am current working on a project in university about speech recognition on Android platform. I need guide on how to overcome "javax" issue.
Regards:
Khurram Shehzad

[Amaury Crickx]

Hi Khurram,

Sorry, but I'm not sure I understand the question. I guess all you want to do is delete the stuff that doesn't compile: in a nutshell, all the File related methods in Recognito.java

Does this answer your question ? 

Cheers,

Amaury

[Amaury Crickx]

Oh, and by the way, Recognito is about speaker recognition, not speech recognition but I guess you already figured out :-)

[Khurram Shehzad]

Hi,
Thanks for quick response.

i already did delete the unwanted stuff, but the problem starts when i modify recognito class "convertFileToDoubleArray()" function.
Now the Normalizer threw the following exception: "Expected value for audio are in the range -1.0 <= v <= 1.0 ".
how do i overcome this. i tried different audio formats but the same result.

Ooops, typing mistake i do know recognito is about speaker recognition not about speech recognition.

[Khurram Shehzad]

here's modified files... "https://www.dropbox.com/s/k2dxohjzemlvat9/Recognito.java?dl=0"

[Amaury Crickx]

Hi,

I didn't try to execute the code but I already see a problem: not all bytes in the file are part of the audio stream.

Here is a description of the WAV file format: https://ccrma.stanford.edu/courses/422/projects/WaveFormat/

Can't tell if the way you're converting the numbers is correct by merely looking at it :-) (also I'm sick and got fever which doesn't help)

In this case, I'd just limit the accepted file format to a single and copy the conversion from FileHelper which is correct.

HTH

Amaury

[Khurram Shehzad]

Hi,
get well soon :))

1-"I didn't try to execute the code but I already see a problem: not all bytes in the file are part of the audio stream."
what i understand is that i have to separate the data part from wav file and then call convert to double[].

2-"In this case, I'd just limit the accepted file format to a single and copy the conversion from FileHelper which is correct."
i didn't get what you meant by format to a single...

[Khurram Shehzad]

Hi,
get well soon :))

1-"I didn't try to execute the code but I already see a problem: not all bytes in the file are part of the audio stream."
correct me if i am wrong, what i get from this is, i have to separate the data part from wav file and then  convert it into double[].

2-"I'd just limit the accepted file format to a single"
i didn't get this, will you please explain what do you meant by format to a single.

On Thursday, October 16, 2014 4:03:42 PM UTC+5, Amaury Crickx wrote:

[Amaury Crickx]

Hi :-)

Already feeling a lot better, thanks !

 

1-"I didn't try to execute the code but I already see a problem: not all bytes in the file are part of the audio stream."
correct me if i am wrong, what i get from this is, i have to separate the data part from wav file and then  convert it into double[].

Yes, reading a particular binary file format (wav, aiff, jpg, pdf, ...) means you have to look at the conventions / rules for the ordering of the bytes and interpret them accordingly.

Now this is quite a tedious thing to do and because it's such a common need, I first looked for libraries that just do this for us on the Android platform.

After an half an hour or so of google searches, I realised it's not something easy to find... 

I guess this could help: http://alvinalexander.com/java/jwarehouse/android/core/java/android/speech/srec/WaveHeader.java.shtml

 

 

2-"I'd just limit the accepted file format to a single"
i didn't get this, will you please explain what do you meant by format to a single.

There are dozens of audio file formats out there. I guess the most known ones are WAV, MP3, AIFF

Wav is usually not compressed and (usually) created with PCM encoding. I say "usually" because it's not required by the Wav specification:

WAV is a so called container format which doesn't enforce the actual encoding of the data, it just structures the data chunks.

Given the myriads of possibilities, I was suggesting that you'd only support e.g.

• Wav
• PCM 
  

With a precision of 

• 16bit 
  
• 16000 Hz 
  
• mono
  

as this is probably the most adequate precision for speaker recognition.

Once you're able to get to the actual sound data, you can reuse the code from FileHelper.java to transform those 16 bits unsigned int values into doubles

HTH

Amaury

[Khurram Shehzad]

Hi,

Thankx for detailed explanation and searching the solution for me. :))

Regards
Khurram Shehzad

[Khurram Shehzad]

Hi,
hopefully you will be good by now.

I have modified the code now its working fine on android, soon i'll share it here after some testing.
Thanks alot it was great help..

Regards
Khurram Shehzad

On Saturday, October 18, 2014 1:13:29 AM UTC+5, Amaury Crickx wrote:

[Giorgos Savvidis]

Hi Khurram,

I'm facing the same problem. Could you please share your code if you have already found a solution that makes recognito Android compatible?

Thanks in advance,

Giorgos.

[yagna.cali...@gmail.com]

Khurram Shehzad 

Hey will u share your android code ?