Description
For Russian speech recognition, generic multilingual models fall short. GigaAM from Sber is an open foundation model for speech recognition built on large-scale self-supervised pretraining.
It ships transcription and emotion recognition models that lead Russian benchmarks, deployable locally.
ASR:Accurate.
Emotion:Recognition.
Pretraining:Self-supervised.
Local:Open weights.
It ships transcription and emotion recognition models that lead Russian benchmarks, deployable locally.
Features
ASR:Accurate.
Emotion:Recognition.
Pretraining:Self-supervised.
Local:Open weights.
