TinyML - Voice on Raspberry Pi Pico.

So I recently bought a Raspberry Pi Pico (RP2040) and I also found this amazing website called edge impulse(www.edgeimpulse.com) which lets us train tinyML models without any code and gives us the library optimized to the microcontroller board we choose.

Now I want to make a tinyML project to do operations through my voice. Raspberry Pi doesn't have a microphone so I have to connect a microphone module to its ADC pin. But I am not able to get how to move forward with this. How can I use the C++ that I export through edge impulse and use it in my code? Does anyone has any experience building similar project? If someone could guide me with how to do this suing tensorFlowLite also that would be helpful.

dougw over 2 years ago

Here is an example - maybe you can have a look at the source code which is on Git Hub... it seems to use TensorFlow...

https://www.youtube.com/watch?v=V0KXZGhHUQY&ab_channel=Forbes
Cancel
Vote Up +3 Vote Down

Sign in to reply

Cancel
feiticeir0 over 2 years ago

Hi !

I've only experience with Python using the Pico and PDM microphone for ML projects.

Nonetheless, there's some examples that use C++

https://blog.tensorflow.org/2021/09/TinyML-Audio-for-everyone.html

https://towardsdatascience.com/audio-deep-learning-made-simple-sound-classification-step-by-step-cebc936bbe5

Coursera has some courses of embedded Machine Learning - using TinyML and Pico or Arduino Sense

https://www.coursera.org/learn/introduction-to-embedded-machine-learning/home/week/1

There are some excellent, excellent books on that topic

https://www.oreilly.com/library/view/tinyml/9781492052036/

https://www.packtpub.com/product/tinyml-cookbook/9781801814973?_ga=2.81539028.270938506.1677456360-1555116499.1677456360

With this, you can create your project.
Cancel
Vote Up +3 Vote Down

Sign in to reply

Cancel
Sudeep AJ over 2 years ago in reply to feiticeir0

Can you please tell me more about your Pico and PDM Microphone ML project using Python?
Cancel
Vote Up 0 Vote Down

Sign in to reply

Cancel
feiticeir0 over 2 years ago in reply to Sudeep AJ

Hi !

You know you can program the Pico with the Arduino IDE using mbed . That way, you can use Edge Impulse with the PICO.

About that project I'm still working on that... Still haven't found a working way to use the PDM microphone with Python. You need to use the ADC pins, but still no luck.

Using the PICO and audio , check this google colab

https://colab.research.google.com/github/ArmDeveloperEcosystem/ml-audio-classifier-example-for-pico/blob/main/ml_audio_classifier_example_for_pico.ipynb

It has everything you need to start your project - it's a perfect example of using a Raspberry PICO with a microphone and do some inference on it.

It goes through all the steps - from data collecting, model creation and training and the steps to do inference on the pico.

The google colab comes from this page

https://blog.tensorflow.org/2021/09/TinyML-Audio-for-everyone.html

Just try it !
Cancel
Vote Up +1 Vote Down

Sign in to reply

Cancel
Sudeep AJ over 2 years ago in reply to feiticeir0

Thank you very much for sharing these with me! I will surely try it.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Cancel
bidrohini over 2 years ago

Can you suggest some machine learning projects that can be done with these tinyML models without using a camera? I have a Raspberry Pi Pico. I want to do ML projects with that. But all the suggestions I get are of camera based projects. I do not have a pico compatible camera.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Cancel
dougw over 2 years ago in reply to bidrohini

https://www.youtube.com/watch?v=V0KXZGhHUQY&ab_channel=ArmSoftwareDevelopers
Cancel
Vote Up +1 Vote Down

Sign in to reply

Cancel
feiticeir0 over 2 years ago in reply to bidrohini

On my comment above, there are sound projects.

Here's one:

https://colab.research.google.com/github/ArmDeveloperEcosystem/ml-audio-classifier-example-for-pico/blob/main/ml_audio_classifier_example_for_pico.ipynb
Cancel
Vote Up 0 Vote Down

Sign in to reply

Cancel