Struct CamPlusPlusExtractor 

pub struct CamPlusPlusExtractor { /* private fields */ }

CAM++ embedder (Channel-Attentive Multi-scale Pooling). Dim is supplied explicitly because WeSpeaker ships several CAM++ variants: voxceleb_CAM++.onnx is 512-d; smaller variants exist at 192-d. Targets the Mobile profile of v1.0; M5 may swap to INT8 + smaller dim. Uses the same 80-bin log-mel fbank pipeline as ResNet34.

Implementations

impl CamPlusPlusExtractor

pub fn new( path: impl AsRef<Path>, dim: usize, pool_size: usize, ) -> Result<Self, EmbedderError>

{ true } pub fn new( path: impl AsRef<Path>, dim: usize, pool_size: usize, ) -> Result<Self, EmbedderError> { ret.as_ref().map_or(true, |e| e.dim() == dim) } Load a CAM++ ONNX model. dim must match the model’s output dimension (e.g. 192 or 512 depending on the variant). Pool size controls the number of concurrent ONNX sessions held internally (canonical: num_cpus().min(4)).

Trait Implementations

impl Embedder for CamPlusPlusExtractor

fn dim(&self) -> usize

Output dimension of this embedder. Constant per instance.

fn embed(&self, audio: &[f32]) -> Result<Vec<f32>, EmbedderError>

Compute an embedding for one audio segment.

fn embed_batch(&self, audios: &[&[f32]]) -> Result<Vec<Vec<f32>>, EmbedderError>

Compute embeddings for a batch of audio segments. Default implementation is sequential; impls may override with a true batched ONNX call.