Blog
Speech data
Speech data
How to Build a Speech Dataset for ASR: Scripts, Recording Specs and QC That Hold Up
IndicVoices asked every speaker for 20 minutes and checked every clip by ear. A method for ASR dataset collection: scripts, consent, specs, QC and WER.
Shivam Singh · 14 min readSpeech data
Code-Switching in Indian Voice Data: What 2,731 Scripts and 805 Typed Answers Show
90% of 2,185 Indian-language reading scripts on our platform mixed in English words, and 47% of Hindi transcriptions were typed in Latin letters. What it means.
Shivam Singh · 13 min readSpeech data
Speech Data Collection Tools Compared: Common Voice Has Just 16 Validated Hours of Hindi
Common Voice holds 29,295 validated hours, but only 16 in Hindi. We compare Common Voice, Karya, Appen, Defined.ai and web forms for collecting speech data.
Shivam Singh · 14 min read