CoolFace
Datasetpublic

dipit099/bengali-tts-yt-test

Bengali TTS — YouTube Pipeline (test) YouTube-sourced Bengali speech chunks transcribed with Gemini 2.5 Flash, produced by the YT scraping + VAD chunking + Gemini transcription pipeline. Columns Column Description uuid Unique identifier (<video_id>_<chunk#>) speaker Speaker / channel name video_id YouTube video ID chunk_file Chunk filename audio_file <speaker>_<video_id>_<chunk_file> duration Duration in seconds transcription Gemini… See the full description on the dataset page: https://huggingface.co/datasets/dipit099/bengali-tts-yt-test.

sourceHugging Facecc-by-4.0updated 4mo agoView on Hugging Face
0likes33downloads
Dataset Card

Bengali TTS — YouTube Pipeline (test)

YouTube-sourced Bengali speech chunks transcribed with Gemini 2.5 Flash, produced by the YT scraping + VAD chunking + Gemini transcription pipeline.

Columns

ColumnDescription
uuidUnique identifier (<video_id>_<chunk#>)
speakerSpeaker / channel name
video_idYouTube video ID
chunk_fileChunk filename
audio_file<speaker>_<video_id>_<chunk_file>
durationDuration in seconds
transcriptionGemini transcription text
audioAudio (WAV, 16-bit PCM, mono, 22050 Hz)
gemini_transcriptionGemini transcription text
whisper_transcriptionWhisper ASR output (NULL — not run yet)
werWord Error Rate (NULL — not computed yet)
cerCharacter Error Rate (NULL — not computed yet)
has_englishEnglish code-switching flag (NULL — not computed yet)
decisionKEEP / EXCLUDE