Re: [INFO] MFCC extraction code using vDSP (iOS/OSX)
Re: [INFO] MFCC extraction code using vDSP (iOS/OSX)
- Subject: Re: [INFO] MFCC extraction code using vDSP (iOS/OSX)
- From: tahome izwah <email@hidden>
- Date: Tue, 19 Apr 2011 17:03:19 +0200
Nice, thank you for sharing this. You might want to know that some of
the projects don't compile though. For instance, SimodOne is missing
the WordMatch library, and once you compile and add it you get
Undefined symbols:
"_WMGetPreProcessInfoForFile", referenced from:
-[SOAppDelegate application:didFinishLaunchingWithOptions:] in
SOAppDelegate.o
-[SOAppDelegate application:didFinishLaunchingWithOptions:] in
SOAppDelegate.o
-[SOAppDelegate recordFileA:] in SOAppDelegate.o
-[SOAppDelegate endEditingBeginThresholdDB:] in SOAppDelegate.o
-[SOAppDelegate endEditingBeginThresholdDB:] in SOAppDelegate.o
-[SOAppDelegate endEditingEndThresholdDB:] in SOAppDelegate.o
-[SOAppDelegate endEditingEndThresholdDB:] in SOAppDelegate.o
-[SOAppDelegate recordFileB:] in SOAppDelegate.o
"_WMGetMinDistanceForFile", referenced from:
-[SOAppDelegate executeComparison:] in SOAppDelegate.o
ld: symbol(s) not found
clang: error: linker command failed with exit code 1 (use -v to see invocation)
Too bad, as I'd really like to play with it.
--th
2011/4/19 Heinrich Fink <email@hidden>:
> Hi,
> I have recently built a robust and efficient implementation to extract
> Mel-Frequency-Cepstral-Coefficients from a window of audio samples. This is
> achieved by using routines provided by the Accelerate.framework only (no
> FFTW dependencies). MFCCs are used widely in the domain of audio analysis,
> especially for speech recognition. As I figure that many readers of this
> mailing list might be working on similar projects, I would like to share
> this code with you:
> https://github.com/hfink/matchbox/blob/master/Xcode/WordMatch/MFCCProcessor.hpp
> https://github.com/hfink/matchbox/blob/master/Xcode/WordMatch/MFCCProcessor.cpp
> I have tested the correctness of this implementation against a commonly used
> MFCC implementation for
> Matlab: http://labrosa.ee.columbia.edu/matlab/rastamat/
> Especially on newer iOS devices equipped with the armv7 NEON SIMD
> instructions, this implementation should be pretty fast!
> This code was developed during a lab course assignment at our university. A
> summary of this project is found here:
> http://www.hfink.eu/matchbox
> Feedback is very welcome, and I hope some of you find the MFCC extraction
> useful! Sources are licensed under the MIT license.
> best regards,
> Heinrich Fink
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Coreaudio-api mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden