speakrs 0.6.0

Fast Rust speaker diarization with pyannote-level accuracy and native CPU/CoreML/CUDA inference
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23

mod data;
mod error;
mod extract;
mod layout;

pub(crate) use data::FrameActivations;
#[cfg(feature = "_metrics")]
#[cfg_attr(docsrs, doc(cfg(feature = "_metrics")))]
pub use data::InferenceStageTimings;
#[cfg(all(feature = "_metrics", feature = "coreml"))]
pub(super) use data::InferenceStageTimings as InternalInferenceStageTimings;
pub use data::{
    BatchInput, ChunkEmbeddings, ChunkSpeakerClusters, DecodedSegmentations, DiarizationResult,
    DiscreteDiarization, InferenceArtifacts, SpeakerCountTrack,
};
pub(super) use data::{
    EmbeddingPath, InferencePath, PendingEmbedding, PendingSplitEmbedding, RawSegmentationWindows,
};
pub use error::PipelineError;
pub(super) use extract::{Array3Writer, EmbeddingStorage, flush_masked, flush_split};
pub(super) use layout::{ChunkLayout, chunk_audio_raw};
#[cfg(test)]
pub(super) use layout::{chunk_start_frames, total_output_frames};