calpoly-csai / swanton

Swanton Pacific Ranch chatbot with a knowledge graph
MIT License
3 stars 1 forks source link

added deepspeech streaming inference #14

Closed chidiewenike closed 4 years ago

chidiewenike commented 4 years ago

Summary

Streams the audio and runs inference in real-time through DeepSpeech.

Details

The function takes in the recording time for the audio with the predefined audio specification constants. The function streams the audio for the input amount of time and runs inference in real-time through DeepSpeech. The function then returns a string of the audio converted to text within the requested time.

Testing

Tested on the deepspeech-0.8.1-models version with both pbmm and tflite formats. Pbmm was testing on a Windows desktop and tflite was tested on an RPi 4.