Author Topic: Can Vamp help me with this?  (Read 3476 times)

beshoo

  • Newbie
  • *
  • Posts: 1
    • View Profile
Can Vamp help me with this?
« on: November 11, 2013, 20:47:29 »
Dear all , i need your help

i'm looking to re recognize "baby crying sound" in order to know the baby needs.
I want to deal with the sound as an event, NOT as music.

I did record a lot of baby crying sounds and labeled each sample with the  "baby status" , is he / she hungry , tired..etc.
in order to teach the system.

Situation :
- I have a lot of labeled WAV files, and we need to insert them "some how" in a database to be able to search for similarity.

User Situation:
- A baby start crying , and we need to identify / analyze the sound/
- Mother will hold a Mobile Device to record a sample of the crying sound. 30 sec.
- The sample will be sent to a "Linux" server to be classified by compare it  with "Our collected database" .
- Server will return the best matched results.

what i need is some think like :
www.findsounds.com/palette.html which provided by http://www.comparisonics.com/api.htm
but it is NOT an open source and they dont provide the API as OpenSource.

Please may you advice me what should i do, and is there any ready to use software with a good accuracy rate ?

Thank you very much.

Vampire

  • Newbie
  • *
  • Posts: 11
    • View Profile
Re: Can Vamp help me with this?
« Reply #1 on: February 04, 2014, 16:36:40 »
you could process the sound - take in buffer by buffer

choose a feature - spectral centroid, mfcc (higher parts)

turn your sample of 30sec into a list of numbers for these features

try machine learning on recognising if the sound is baby/not baby

so could vamp help? - probably - you could try taking some of the routines here and using them to get your features. You probably don't want to buld a vamp host or anything crazy. just turn audio into a feature set that you can learn on.

hope that helps a bit! I don't necessarily know how to do all the above.