picovoice 0.8.4

Picovoice is an end-to-end platform for building voice products on your terms.

Picovoice Python SDK

Picovoice is an end-to-end platform for building voice products on your terms. It enables creating voice experiences similar to Alexa and Google. But it entirely runs on-device. Picovoice is

• Private: Everything is processed on-device. Intrinsically HIPAA and GDPR compliant.
• Reliable: Runs without relying on connectivity or external service availability.
• Responsive: Edge-first architecture eliminates unpredictable network delay.
• Highly-Accurate: Resilient to noise and reverberation. Outperforms cloud-based solutions with high margins.
• Cross-Platform: Build on your platforms of choice, with tools your team is accustomed to.
• Cost-Effective: Avoid unbounded cloud fees by processing locally with minimal resources.

Compatibility

• Python 3
• Runs on Linux (x86_64), Mac (x86_64), Windows (x86_64), Raspberry Pi (all variants), and BeagleBone.

Installation

pip3 install picovoice

Usage

Create a new instance of Picovoice platform as below. keyword_path is the absolute path to Porcupine Wake Word Engine keyword file (with .ppn suffix). context_path is the absolute path to [Rhino Speech-to-Intent Engine] context file (with .rhn suffix). wake_word_callback is invoked upon the detection of wake phrase and inference_callback is called upon completion of follow-on voice command inference.

from picovoice import Picovoice

keyword_path = ...

def wake_word_callback():
    pass

context_path = ...

def inference_callback(is_understood, intent, slot_values):
    pass

pv = Picovoice(
        keyword_path=keyword_path,
        wake_word_callback=wake_word_callback,
        context_path=context_path,
        inference_callback=inference_callback)

When instantiated, valid sample rate can be obtained via pv.sample_rate. Expected number of audio samples per frame is pv.frame_length. The incoming audio can be processed as below

def get_next_audio_frame():
    pass

while True:
    pv.process()

Finally, when done be sure to explicitly release the resources.

pv.delete()

References

• File Demo
• Microphone Demo