• State TI Thinks Resolved
  • Locked Locked
  • Replies 5 replies
  • Answers 1 answer
  • Subscribers 31 subscribers
  • Views 1163 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.

Suggestion for an audio and internet based project

Caner Alp
Prodigy 215 points

Other Parts Discussed in Thread: ALP, TMS320C5517

Hello,

Im a student and I have a project which converts speech to text,  but I want it to be done over internet by using a service ( such as IBM's speech to text service).

Here is a question. is it possible to do this with an TM4C microcontoller, it will record sound and save as a suitable format for the service, then send it to the cloud and get the data (words). Like IOT devices but here the datas are bigger.

Or if you have any suggestions, please help me.  I may use a DSP as well. 

Thank you.

Caner ALP

  • 0
    TI__Guru 59540 points
    Hello Caner,

    Unfortunately there isn't enough information provided to give a complete answer.

    First, how are you planning on storing the recorded information or are you planning on live streaming it to the cloud? Certainly TM4C would have the capability to interface to an external storage device such as an SD card or other flash based device but TM4C would probably not have enough storage capability to record very much spoken word.

    Second, how do you plan on converting the spoken word into the digital format you need for IBMs cloud? Perhaps through ADC inputs? Not knowing the format is a major obstacle to deciding what would need to be done by the CPU and if it could keep up with the needed sampling rate.

    I certainly think that the TM4C together with some external circuitry and peripheral devices could do the job, I simply don't know how effective it would be given the performance available. At least, it would be a great learning opportunity to do this using the CIMSIS DSP library running on an M4F device .

    For a greater probability of success, however, I would recommend having a look at one the beaglebone black development kits with a Sitara class device on it (www.ti.com/.../beaglebk black&tisearch=Search-EN-Everything). I think they may already have support for IBM cloud connectivity as well.
  • 0 in reply to Chuck Davenport
    Prodigy 215 points
    Thank you for replying,

    Im thinking to use TMS320C5517 DSP, I will use a codec processor as well, i need to record the sound as wav, then live streaming to the cloud.
    I will add RAM and ROM also. If I use TIRTOS, cant I do this ?

    "The IBM service is accessed via a WebSocket connection or REST API.
    To run example cURL commands that work with the service from the command line, see Quick start for cURL.
    To create and run an example Node.js application that works with the service from the command line, see Watson Quick Start for Node.js.
    To create and run a sample Node.js application that works with the service from a web browser, see Developing a Watson application in Node.js. You need the link to the source code for the Node.js application at the speech-to-text-nodejs repository in the watson-developer-cloud namespace on GitHub." this is written in IBM's webpage.

    I dont have knowledge about internet programming, that is why I need help.

    Since My aim is to make a wearable device, I should use small ICs and low power.

    Best Regards

    Caner Alp
  • 0 in reply to Chuck Davenport
    Prodigy 215 points
    Thanks for replying,

    Actually my project is a wearable device which converts speech to text in real time for deaf people. If I can save sound as wav and send it to the cloud ,I will be able to do this job. But there is something not clear how to send the data to the colud.

    Ive made some research, Im planning to use a DSP ( TMS320C5517 ), codec, RAM - ROM , and screen, since this is a wearable project,everything must be very low power. And If run TIRTOS on it, can I do this. I will paste the description of IBM's service, I dont have knowledge about internet programming, that's why it comes to me diffucult.

    "To run example cURL commands that work with the service from the command line, see Quick start for cURL.
    To create and run an example Node.js application that works with the service from the command line, see Watson Quick Start for Node.js.
    To create and run a sample Node.js application that works with the service from a web browser, see Developing a Watson application in Node.js. You need the link to the source code for the Node.js application at the speech-to-text-nodejs repository in the watson-developer-cloud namespace on GitHub."

    "The service is accessed via a WebSocket connection or REST API."

    As it is written, I should use cURL and Node.js, Can I make it compatible with my Embedded system this is what my question is actually.

    Best Regards

    Caner Alp
  • 0 in reply to Caner Alp
    TI__Guru 59540 points
    Hello Caner,

    Unfortunately I have no experience implementing an IBM cloud interface. Our TM4C129 launchpads come with an interface to the Exosite cloud but this would be completely different than IBM. I am certain there is probably an SDK from IBM that would be needed to support their services and I don't know if this would include anything related to their voice to text support. I would suggest you check in with the forum of the DSP you are planning on using and see if anyone has done anything with the IBM cloud and has experience with the sort of thing you are trying to accomplish. For now, I don't think I would be of much additional help.

    You may also want to check in our TI Design repository at www.ti.com\tidesigns to see if there are any existing designs that would be of help in your project.
  • 0 in reply to Chuck Davenport
    Guru 117855 points

    Chuck Davenport said:
    Unfortunately there isn't enough information provided to give a complete answer.

     Might it be that simple, "misfortune" stands (not) alone in the creation of, "Insufficient information?"   (i.e. has a partner in crime!)

    For years here - others & myself have made the case for, "Poster Guidelines" which enforce a structure & discipline upon posters so that, "Measurably better" posts result.   Buried far back w/in this forum are "working models" of such Poster Guidelines - which have (always) been (duly) noted & rejected...

    You & other vendor staff surely try - yet the expectation that (most) posters are adequately skilled, disciplined, & caring seems more "wishful" than real - and time, effort & clarity are sacrificed.   That's the real, "misfortune" and it's permitted to go on & on.   (and "that's" duly noted!)