Russian Speech Data With Pitch and SNR Annotations

Annotated Russian audio data for tasks like text-to-speech, speech recognition, and speaker identification. The dataset includes features such as text transcriptions, speaker identifiers, audio files, utterance pitch statistics, and signal-to-noise ratio (SNR). It was created by author kijjjj and last updated in June 2025.

Use Cases

• Train a speech recognition model using the text transcriptions and corresponding audio files.
• Develop a speaker identification system by clustering or classifying audio based on the speaker_name field.
• Analyze speech prosody and intonation patterns using the utterance_pitch_mean and utterance_pitch_std features.
• Filter or augment training data for text-to-speech models based on the snr metric to select high-quality audio samples.

Strengths

• Includes multiple analytically rich features like utterance_pitch_mean, utterance_pitch_std, and snr.
• Dataset is categorized as containing between 100K and 1M samples, indicating a substantial size.
• Specifically designed for multiple speech-related tasks, including text-to-speech and speaker identification.

Limitations

• Exact row count, column count, and file size are unknown, making it difficult to assess computational requirements.
• Geographic and demographic coverage of the Russian speakers is unspecified, which may introduce bias.
• The dataset's license is not explicitly stated in the provided input, creating potential usage uncertainty.

Provenance

Source

huggingface, author kijjjj

Freshness

Last updated on 2025-06-26.

Geography

Region is tagged as 'us', but the audio language is Russian.