This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

VoLIB/Codec Server Integration

Other Parts Discussed in Thread: OMAP3530

We have a need to integrate algorithms from the VoLIB into our existing codec server (for the OMAP3530/C64+). Our server currently contains a number of audio & video encoders and decoders, as well as a custom video processing algorithm we wrote in-house that uses DSPLIB 2.1, IMGLIB 2.01, and VLIB 2.1.  I had considered creating a custom audio processing algorithm in the same way as I did the custom video algorithm, but using VoLIB instead of IMGLIB and VLIB. To do this, will I need to use the ECO interface in the VoLIB? I noticed there is also an XDAIS interface. If I can, I think I'd prefer to simply integrate the existing XDAIS interfaces from VoLIB directly into my server so I don't have to create a custom audio algorithm in the server.  Does anyone have any comment on this? Are there any documentation resources (wiki's, etc.) for the VoLIB, besides the included VOLIB_Users_Manual.chm?

Thanks,

Dennis

  • Hi Dennis,

    First of all, I'd like to be sure that VoLIB is what you're looking for.  VoLIB offers high-performance speech processing building blocks.  All components in VoLIB support 8kHz sampling, though some offer configuration for 16kHz as well.  If by "audio" processing you mean 8kHz speech -- you're in business.

    As for the ECO API versus XDAIS -- all components in VoLIB support the ECO API, while only a handful support XDAIS.  So it appears you'll have to support the ECO API unless you only need those components that support the XDAIS API.

    At this point we don't have an external Wiki for VoLIB.  If you have specific questions we will do our best to provide timely answers or direct you to the proper section in the documentation.  This can also help us identify potential areas within the documentation that may need improvement.

    Regards,

    Charlie

     

     

  • Thanks for the reply,

    Our purpose for looking at this library is to process the audio to detect the presence or absence of speech. 8kHz is probably fine.  I know the VAU component supports XDAIS. Does it make more sense to try to export this component directly by our codec server or to write a "wrapper codec" which calls the ECO API internally to provide the needed functionality?

     

    Thanks,

    Dennis

  • If the VAU is the only component you need, then I would just use the XDAIS API that it provides and not bother with any ECO API wrapper.  Also, the VAU supports 16kHz if this would be preferable. 

    If processing resources (MIPS) are tight, another option is the simple VAD (SVD) module of the voice processing elements (VPE) package.  This is a "lightweight" and simple voice activity detection algorithm that works surprisingly well.  This does not, however, support the XDAIS API.

    Regards,

    Charlie