What is a frame?
A frame in audio is a short, windowed segment of the waveform (typically 10–25 milliseconds) used as the basic unit of processing in DSP and voice AI.
What is an example of frame in use?
An ASR system running at 16 kHz typically processes 20 ms frames (320 samples each), computing features like mel-spectrograms on each frame as input to the acoustic model.
How does ai-coustics use frames?
ai-coustics models are frame-based and built for streaming: they process short audio frames with minimal lookahead, which is what enables ultra-low latency. Combined with our AirTen inference engine, this lets enhancement run in real time on CPU.
