Expand description
§Kalosm Sound
Kalosm Sound is a collection of audio models and utilities for the Kalosm framework. It supports several voice activity detection models, and provides utilities for transcribing audio into text.
§Sound Streams
Models in kalosm sound work with any AsyncSource
. You can use MicInput::stream
to stream audio from the microphone, or any synchronous audio source that implements rodio::Source
like a mp3 or wav file.
You can transform the audio streams with:
VoiceActivityDetectorExt::voice_activity_stream
: Detect voice activity in the audio dataDenoisedExt::denoise_and_detect_voice_activity
: Denoise the audio data and detect voice activityAsyncSourceTranscribeExt::transcribe
: Chunk an audio stream based on voice activity and then transcribe the chunked audio dataVoiceActivityStreamExt::rechunk_voice_activity
: Chunk an audio stream based on voice activityVoiceActivityStreamExt::filter_voice_activity
: Filter chunks of audio data based on voice activityTranscribeChunkedAudioStreamExt::transcribe
: Transcribe a chunked audio stream
§Voice Activity Detection
VAD models are used to detect when a speaker is speaking in a given audio stream. The simplest way to use a VAD model is to create an audio stream and call VoiceActivityDetectorExt::voice_activity_stream
to stream audio chunks that are actively being spoken:
use kalosm::sound::*;
#[tokio::main]
async fn main() {
// Get the default microphone input
let mic = MicInput::default();
// Stream the audio from the microphone
let stream = mic.stream().unwrap();
// Detect voice activity in the audio stream
let mut vad = stream.voice_activity_stream();
while let Some(input) = vad.next().await {
println!("Probability: {}", input.probability);
}
}
Kalosm also provides VoiceActivityStreamExt::rechunk_voice_activity
to collect chunks of consecutive audio samples with a high vad probability. This can be useful for applications like speech recognition where context between consecutive audio samples is important.
use kalosm::sound::*;
use rodio::Source;
#[tokio::main]
async fn main() {
// Get the default microphone input
let mic = MicInput::default();
// Stream the audio from the microphone
let stream = mic.stream().unwrap();
// Chunk the audio into chunks of speech
let vad = stream.voice_activity_stream();
let mut audio_chunks = vad.rechunk_voice_activity();
// Print the chunks as they are streamed in
while let Some(input) = audio_chunks.next().await {
println!("New voice activity chunk with duration {:?}", input.total_duration());
}
}
§Transcription
You can use the Whisper
model to transcribe audio into text. Kalosm can transcribe any AsyncSource
into a transcription stream with the AsyncSourceTranscribeExt::transcribe
method:
use kalosm::sound::*;
#[tokio::main]
async fn main() {
// Get the default microphone input
let mic = MicInput::default();
// Stream the audio from the microphone
let stream = mic.stream().unwrap();
// Transcribe the audio into text with the default Whisper model
let mut transcribe = stream.transcribe(Whisper::new().await.unwrap());
// Print the text as it is streamed in
transcribe.to_std_out().await.unwrap();
}
Re-exports§
Structs§
- A stream of
SamplesBuffer
s with voice activity detection information - A microphone input.
- A stream of audio data from the microphone.
- Error that reports the unsupported value
- Error that reports the unsupported value
- A resampled async audio source
- A transcribed segment of audio.
- The output of a
crate::VoiceActivityDetectorStream
- A stream of
SamplesBuffer
s with voice activity detection information - A stream of audio chunks that have a voice activity probability above a given threshold
- A stream of audio chunks with a voice activity probability rolling average above a given threshold
- A quantized whisper audio transcription model.
- A builder with configuration for a Whisper model.
Enums§
- The progress starting a model
- A language whisper can use
- The source whisper model to use.
Traits§
- A streaming audio source for single channel audio. This trait is implemented for all types that implement
rodio::Source
automatically. - An extension trait for
AsyncSource
that integrates withcrate::Whisper
. - An extension trait for audio streams for denoising. Based on the nnnoiseless crate.
- An extension trait to transcribe pre-chunked audio streams
- An extension trait for audio streams that adds a voice activity detection information. Based on the voice_activity_detector crate.
- An extension trait for audio streams with voice activity detection information