• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Subscribers 100 subscribers
  • Views 862 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

Implementation of speech enhancement algorithm from SIMULINK using embedded target toolbox on DSK6713

jigar shah
Prodigy 40 points

I have developed and simulated a speech enhancement algorithm which uses a combination of MMSE (minimum mean square error) STSA (short time spectral attenuation) and RASTA (relative spectral amplitude) approaches.

    It gives noise reduction in various noise conditions like airport, car, train, station, babble, restaurant etc. upto 0dB SNR.
    The simulation in MATLAB/ SIMULINK works well on my laptop and desktop (with soundcard provides input-output speech).
    But same when I download on TMS320C6713 DSP using DSK 6713; the original noise is reduced but some continuous          background noise is introduced which was not present when I simulated same on laptop/desktop.

    Can you guide me about the possible reasons for this problem?
    This will help me a lot to progress further in my research.
    Can I change the implementation platform from DSK6713 to Media processor DM6437? Will it eliminate my problem?
-Jigar H. Shah
  • 0
    TI__Expert 4680 points

    Changing platform won't change the noise issue. The processor core will for DSP optimisation will work the same. The issue will be on the algorithm or how you are taking the input and output. I am assuming with simulator there was no physical input it is only the file input.

    If you are taking live input then check your mike used for input. What OS you are using for your devlopment and how are you accessing the peripherals.

    Regards

    Asheesh

  • 0 in reply to Asheesh Bhardwaj
    Prodigy 40 points

    THANKS.

    The input is given from Line out of PC(the speech file is played on windows media player continuously) to the ADC of DSK6713. So no question of problem with microphone.

    On SIMULINK the speech is through file input as well as through Mic In of sound card on PC which is driven by Line out from CDPlayer. The file input and live input both are giving same and nice result on SIMULINK. Same algorithm with the same configuration of Input-Output gives noisy speech output when downloaded on DSK 6713.

    If problem exists in algorithm the SIMULINK output must be noisy which is not the case?

    Can you elaborate further?

    with regards

    -Jigar Shah

  • 0 in reply to jigar shah
    TI__Expert 8150 points

    I recommend you ask this query here: http://www.mathworks.com/matlabcentral/newsreader/ 

    It will be difficult for us to understand the reason of the noise as the code generation for the target is handled by mathworks tools. 

  • 0
    Prodigy 10 points

    Sir,

    i am PG student. I am doing my research in the field of speech enhancement. It would be helpful for me if you provide your simulink model and your project report.

    my mail id: ernandeshwaran@gmail.com