# Golos: Russian Dataset for Speech Research

arXiv: https://arxiv.org/abs/2106.10161 · HTML: https://mlsanity.com/paper/2106.10161/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 18 Jun 2021.

## Built on by (most cited first, 9 of 9)

- [LibriSpeech-PC: Benchmark for Evaluation of Punctuation and Capitalization Capabilities of end-to-end ASR Models](https://mlsanity.com/paper/2310.02943/) (2023)
- [MSR-86K: An Evolving, Multilingual Corpus with 86,300 Hours of Transcribed Audio for Speech Recognition Research](https://mlsanity.com/paper/2406.18301/) (2024)
- [MOLE: Metadata Extraction and Validation in Scientific Papers Using LLMs](https://mlsanity.com/paper/2505.19800/) (2025)
- [Towards Robust Speech Deepfake Detection via Human-Inspired Reasoning](https://mlsanity.com/paper/2603.10725/) (2026)
- [Methods to Increase the Amount of Data for Speech Recognition for Low Resource Languages](https://mlsanity.com/paper/2501.14788/) (2025)
- [TEDI: Trustworthy and Ethical Dataset Indicators to Analyze and Compare Dataset Documentation](https://mlsanity.com/paper/2505.17841/) (2025)
- [Balalaika: Data-Centric, Prosody-Aware Annotation Pipeline for Russian Speech](https://mlsanity.com/paper/2507.13563/) (2025)
- [Pisets: A Robust Speech Recognition System for Lectures and Interviews](https://mlsanity.com/paper/2601.18415/) (2026)
- [GigaAM Multilingual: Foundation Model for Underrepresented Languages](https://mlsanity.com/paper/2607.10371/) (2026)

## Abstract (arXiv)

This paper introduces a novel Russian speech dataset called Golos, a large corpus suitable for speech research. The dataset mainly consists of recorded audio files manually annotated on the crowd-sourcing platform. The total duration of the audio is about 1240 hours. We have made the corpus freely available to download, along with the acoustic model with CTC loss prepared on this corpus. Additionally, transfer learning was applied to improve the performance of the acoustic model. In order to evaluate the quality of the dataset with the beam-search algorithm, we have built a 3-gram language model on the open Common Crawl dataset. The total word error rate (WER) metrics turned out to be about 3.3% and 11.5%.