SparkFun Forums

[d3mac123]

I am trying to have a speech-to-talk feature with my Artemis Nano. I have tried the Talkie library but it is incompatible with Apollo3-based boards. Any suggestions on how to solve my issue?

[jremington]

To adapt the Talkie library to other processors, you will need modify the specific hardware features (in this case, timer usage) to suit.

On AVR based Arduinos, the library uses a timer to generate 8-bit PWM signals that are in turn smoothed to produce analog audio signals, but a reasonably fast DAC will work even better.

[Anthony78]

Hello,
I have some write this you can read this topic
If you are looking for a speech-to-text feature with your Artemis Nano and the Talkie library is not compatible with Apollo3-based boards, there are alternative solutions you can explore. Here are some suggestions to help you achieve your goal:

Use Speech Recognition APIs: Instead of relying on the Talkie library for speech-to-text, you can utilize cloud-based speech recognition APIs from platforms like Google Cloud Speech-to-Text or IBM Watson Speech to Text. These APIs allow you to send audio data from your Artemis Nano to the cloud for speech recognition and receive the corresponding text output.

Implement Local Speech Recognition: Depending on the capabilities of your Artemis Nano and available libraries, you might be able to find local speech recognition libraries or models that run directly on the board. These solutions can be more privacy-focused as they don't rely on sending data to external servers.

Custom Model Integration: If you have expertise in machine learning and natural language processing, you can train your custom speech recognition model and integrate it into your Artemis Nano. Platforms like TensorFlow or PyTorch can be used for creating and deploying custom models.

Before proceeding with any implementation, make sure to review the technical documentation and specifications of your Artemis Nano board to ensure compatibility with the selected solution. Additionally, test the solution thoroughly to validate its accuracy and performance.

If you face any specific technical challenges during the implementation, you may reach out to the hardware's manufacturer or relevant developer communities for guidance and support.
Thanks for read this information, i hope my information is good and helpful.

[jremington]

Deeply grateful to chatGPT for the above useless post.

[YellowDog]

jremington for the win!