yandexdataschool/speech_course

Covers practical speech processing tasks spanning DSP fundamentals, discriminative models (VAD, SED, keyword spotting, speaker verification), automatic speech recognition with CTC/RNN-T decoding and streaming inference, text-to-speech synthesis with acoustic modeling and neural vocoders, and signal enhancement (noise reduction, beamforming). Emphasizes hands-on implementation through weekly seminars and graded assignments, with code examples for mel-spectrogram extraction, contrastive loss training (ECAPA-TDNN), and real-time ASR pipelines.

319 stars.