PocketSphinx currently has no mechanism to align a known transcription with audio to obtain per-word…
PocketSphinx currently has no mechanism to align a known transcription with audio to obtain per-word…: a task in LegoFlow-SWE (Harbor dataset). Expected behaviour: For each utterance ID from the control file, the batch process should read the corresponding transcription file at <aligndir /<uttid…
The task
**Expected behaviour:** For each utterance ID from the control file, the batch process should read the corresponding transcription file at `<aligndir>/<uttid><alignext>` (a single line of whitespace‑separated words), perform force alignment using that transcription, and – when `-hypseg` is specified – output per‑word…
Part of Lego-X/LegoFlow-SWE.