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.

Convert PCM Data to wav

Other Parts Discussed in Thread: CC3200AUDBOOST, CC3200

Dear Community,

I have CC3200 board and CC3200AUDBOOST it has TLV320AIC3254, I am able to convert audio from mic to raw pcm data,

I want to convert raw pcm to wav or flac or mp3 so that i can send it to google for recognition 

Can you please help me

Regards

Shashank

  • Hi, Shashank,

    Welcome to E2E, Thanks for your interest in our products!.

    Unfortunately, We don't have an IC capable of converting PCM data to a different format, you might need a processor with enough capabilities of doing so or an encoding software.

    Best Regards,

      -Diego Meléndez López
       Audio Applications Engineer

  • Hi Diego,

    We are very new to Audio Processing, We are planning to implement Voice Recognition to one of our products

    The goal of this project is that our product will have a mic and will be continuously listening to the stream looking for a keyword ex: Ok Google

    Once the keyword is detected it should start recording the data from mic and send to the google servers for conversion from speech to text

    Kindly help us in this aspect , How should we get started ? What processors or codecs should we be looking at ?

    Regards
    Shashank
  • Hello Leonard,

    Thank you I have gone through the links, Today we have decided that Voice processing should be done Offline

    The C5535 eZdsp uses TIESR gforge.ti.com/.../ . We tried to register in this website for access but we never got any activation emails from them. Is this project Dead ?

    Also if we use this in our project and assume I need 50 Individual words to be stored how much memory is required ? Is there a way to calculate ?

    Kindly help us get access to TIESR asap

    Regards
    Shashank
  • Hi Shashank,

      I see the software is directly downloadable from this Product page: http://www.ti.com/tool/TIDEP0066 .

    There is a Design Guide and other reference material here as well.

    ~Leonard   

  • Hello Leonard,

    Thanks, I have downloaded and i will go through

    Does this DSP support Over the Air updates ? How much memory can be interfaced with it ?

    Regards
    Shashank
  • Shashank,

    As Leonard has mentioned, please take a look at the TIesr TI design. Have you decided on a DSP? The TI design is based on the C5535 ezdsp, but it can also work on the C5545 DSP. C5000 has an e2 forum e2e.ti.com/.../109 for followup questions.

    Please keep in mind that the TIesr TI design is for demo purposes only and not meant to be used in a production system. We do work with 3rd parties that can provide a much more robust voice recognition library. When ready to cross that bridge, please reach out to your TI sales rep and they can relay more info.

    Hope this gets you going. Thanks.

    Lali
  • Hi Lalindra,

    We havent decided on a DSP yet, Can you let us know who are these 3rd parties who can provide robust voice recognition library ? How can we get in touch with them ?

    We will use the DSP which ever suits the voice recognition library, What do you say ? We need around 100 words to be recognized Offline without internet

    Regards
    Shashank
  • Dear TI team ,

    Any update on the above ?

    Regards

    Shashank

  • Shashank,

    We have not tested more than 2 words on TIesr running on the C5535 dsp, so its hard to say for certain that this DSP would be a good fit. 

    Some things to consider in this case is the space that's needed on the device when the vocabulary is expanded to accommodate more words. Also, MIPS overhead with this large library. You also want acoustically unique words for all 100 words so that you avoid several sounding alike.

    I would suggested getting a C5535 eZdsp, adding more words to TIesr and evaluate performance as you scale the vocabulary.

    Is there a TI rep or FAE that you are working with already?

    Lali