ToneWebinars Balalaika: 248 Hours of Annotated Russian Podcast Speech

ToneWebinars Balalaika is a 248.9-hour Russian speech corpus curated from podcasts by the MTUCI lab260 team. Released in early 2026, the dataset was processed using the BALALAIKA pipeline to provide high-quality audio for generative speech tasks. It serves as a refined version of the original ToneWebinars source, specifically filtered for speech synthesis and recognition.

Use Cases

• Training Russian text-to-speech (TTS) models using the BALALAIKA-filtered audio segments
• Improving automatic speech recognition (ASR) for conversational Russian podcasts
• Researching speech generative tasks using the provided transcriptions and audio pairs

Strengths

• 248.9 hours of total audio duration
• Processed via the BALALAIKA pipeline for quality filtering
• Apache 2.0 license allows for open research and commercial use

Limitations

• Restricted to the Russian language
• Genre-specific bias toward podcast-style conversational speech
• Potential for residual background noise inherent in original podcast recordings

Provenance

Source

MTUCI lab260, derived from ToneWebinars

Collection Method

Annotated and filtered using the BALALAIKA pipeline

Freshness

Last updated March 2026.

Geography

Russia