C++ implementation of DistilHuBERT using Eigen.
hubert::model net;
float audio[16000]; // 16kHz mono audio
std::span<const float> feats = net.encode(audio); // [T,768] : an array of 768 packed normalized features- Dynamic sizes (but no batches)
- Weights compiled into the library
- Block-based API
- Streaming API.
- Useful for downstream tasks such as speech representation, speaker identification, classification, etc.