Russian speech database

I. Product Overview

A Russian speech database, totaling 1,279 hours and 517,548 sentences. The sampling rate is 16,000 Hz, with a bit depth of 16 bits. A total of 939 speakers were invited to record using their mobile phones in quiet environments. The recordings were transcribed and annotated by professional linguists and underwent rigorous quality assurance, achieving a text accuracy rate exceeding 95%.

II. Speech Data Transcription

Transcribers reproduce the audio content as accurately as possible, ensuring that the written text matches the spoken content. General guidelines are as follows: 1) The transcription must exactly match the audio; no extra words, missing words, or errors are permitted. 2) Numbers should be transcribed in English Latin form—for example, “raz dva tri” rather than “123.” 3) English words should be written in lowercase, while phonetic spellings must use uppercase letters separated by spaces, such as “W W W dot Google dot com” or “C E O.” 4) All annotations must fully reflect the original pronunciation without any omissions.

III. Directory Structure and Naming Conventions

Directory Structure

The data directory tree is structured as follows:

└─DOC (textual documentation)
││├─wav_list.txt (audio file list)
││├─content.txt (transcription list)
└─SPEECHDATA (audio folder)
││├─F0013 (speaker ID)
││ ││ F0013W0001.wav (audio file)

Naming Conventions

CORPUS/USAGE/SPEAKER_ID/SPEECH_ID e.g., /SPEECHDATA/F0013/F0013W0001.wav

Directory Name · Content · Notes

CORPUS · · Speech database identifier

USAGE · SPEECHDATA · Audio folder

SPEAKER_ID · F0013 · Speaker ID

SPEECH_ID · F0013W0001.wav · Audio file

Data Sample Download

Click to download a data sample