Sign in to view source links and access this dataset
Description
A cleaned and organized Parquet version of the bengali-tts-combined dataset, structured by speaker folders. The dataset includes audio chunks, their durations, and corresponding Bengali transcriptions, sourced from multiple speakers and videos. It was created by author 'smam' and last updated on 2026-06-15.
Use Cases
Train Bengali text-to-speech models based on speaker-specific audio and transcription pairs.
Benchmark speech synthesis quality based on audio duration and transcription accuracy.
Perform speaker adaptation or multi-speaker TTS training based on the 'speaker' folder structure.
Analyze phonetic or prosodic features in Bengali speech based on aligned audio and text data.
Strengths
Data is cleaned and organized into a structured Parquet format.
Includes speaker separation, which can support multi-speaker model development.
Contains both audio file paths and corresponding transcriptions, providing aligned data for TTS.
Limitations
Row count and total dataset scale are unknown, which may limit suitability assessment.
Column-level documentation is absent; field semantics must be inferred after download.
Freshness should be verified as the last update timestamp is 2026-06-15.
Provenance
Source
huggingface, author smam
Collection Method
Derived from the 'bengali-tts-combined' dataset; cleaning and organization method unspecified.
Freshness
Last updated 2026-06-15 06:37:18.
License is unknown; terms of use must be verified before application.