• State
  • Locked Locked
  • Replies 2 replies
  • Subscribers 101 subscribers
  • Views 461 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.

SK-AM62-LP: Audio AI processing guidance needed

bobba beng
Prodigy 10 points

Part Number: SK-AM62-LP

Hey there,

We are wishing to run advanced audio processing using AI on the TI chip in a low power environment. We've done some basic research and feel the AM62s probably fits our use case.

Before when writing (non AI) signal processing algorithms, we would deploy directly onto the 'baremetal' via the cortexM4.

However, all the examples I've seen for the chip/dev board, are using a linux application layer. Our samples load and work on the linux application layer, but I was assuming

we would need to run within the signal processing module.

Generally, we need to:

Access audio inputs/outputs (3.5mm jacks) using I2C and DA/ AD converters

Process chunks of audio (10ms) using ML processing and the full resources on the chip to allow lowest possible latency

Provide means to 'flash' code via USB

Any help and guidance much appreciated. We're just not sure whats the appropriate way to use what is available on the chip.

Thanks

  • 0
    TI__Expert 5380 points

    Hello,

    It is our pleasure to help. Your question is redirected to the expert on the matter. Please, allow some time for them to answer. 

    Best regards,

    Qutaiba

  • 0
    TI__Guru** 116950 points

    Am62 device is capable of this solution however the SOC always requires DDR memory as we have only KB worth of memory and generally need atleast a 256MB DDR memory for running HLOS and AI solution for any modality. You can connect audio to the SOC using MCASP which is capable of supporting upto 32 I2S channels using external ADC/DAC

    What AI OSRT or model do you plan to run. Is the plan to use tinyML or TFlite Micro? Can you indicate why you are considering using SoC with CortexA cores if you only need M4 performance to run Audio ML ?

    Thanks and Regards,

    Rahul