alphacep/awesome-russian-speech

What it solves

Finding reliable tools, datasets, and models for Russian speech processing is scattered and often buried in Russian-language sources. This project gathers them into one organized reference so you can quickly locate what you need without endless searching.

How it works

It is a manually curated list of links, arranged into categories like speech recognition, speech synthesis, datasets, text normalization, punctuation, stress marking, and more. Each category points to specific GitHub repositories, Hugging Face models, datasets, courses, and community channels. The list is maintained by the alphacep team and includes both well-known tools (Vosk, Whisper, Golos) and niche community resources.

Who it's for

Developers, researchers, and students working with Russian speech technologies—whether you are building ASR, TTS, or language preprocessing pipelines. Also useful for anyone exploring the Russian speech ecosystem, including those who prefer resources in Russian or English.

Highlights

• Speech recognition models: Vosk, Whisper, Citrinet, and more, with quality comparisons.
• Speech synthesis tools: Vosk-TTS, RHVoice, TeraTTS, and fine-tuned F5-TTS variants.
• Datasets: Golos, Common Voice, Russian Librispeech, plus many Hugging Face corpora.
• Text normalization and preprocessing utilities for TTS.
• Stress marking, phonetic dictionaries, and grapheme-to-phoneme (g2p) tools.
• Emotion recognition models and datasets.
• Punctuation and capitalization restoration models.
• Courses, community links, and historical background on Russian speech science.