# RUSLAN: Russian Spoken Language Corpus for Speech Synthesis

## TL;DR

RUSLAN is the largest annotated Russian speech corpus for a single speaker, enabling improved text-to-speech synthesis with high naturalness and intelligibility, demonstrated through neural network training and MOS evaluation.

## Contribution

The paper introduces RUSLAN, the largest Russian speech corpus for a single speaker, and demonstrates its effectiveness for neural TTS with quality evaluation.

## Key findings

- Synthesized speech scored 4.05 for naturalness.
- Synthesized speech scored 3.78 for intelligibility.
- Corpus contains over 31 hours of high-quality speech.

## Abstract

We present RUSLAN -- a new open Russian spoken language corpus for the text-to-speech task. RUSLAN contains 22200 audio samples with text annotations -- more than 31 hours of high-quality speech of one person -- being the largest annotated Russian corpus in terms of speech duration for a single speaker. We trained an end-to-end neural network for the text-to-speech task on our corpus and evaluated the quality of the synthesized speech using Mean Opinion Score test. Synthesized speech achieves 4.05 score for naturalness and 3.78 score for intelligibility on a 5-point MOS scale.