Per-Word Speech Transcript Timestamping

via Freelancer ·

Budget / SalaryHourly project
TypeFreelance project
LocationRemote
Posted2 hours ago
I have a collection of raw audio recordings that already come with draft transcripts. I need each spoken word in those transcripts precisely time-stamped so the text aligns perfectly with the audio.

Scope of work
• Load the supplied audio files and their matching text.
• Verify the wording quickly so the transcript still mirrors the speech (minor typos are fine to fix as you go).
• Insert a start-and-end timestamp for every single word—sentence-level timing is not enough.
• Export the finished annotation in either TextGrid, ELAN, or a clean CSV/JSON format—whichever you are most comfortable using.

Quality expectations
• Word timing must be consistently within ±50 ms of the actual audio.
• No extra tags are required (speaker labels, emotion cues, background-noise markers, etc.).
• Deliver files in the same folder structure I provide so they can be batch-processed by my downstream tools.

If you have previous experience aligning transcripts, building ASR corpora, or using tools like Praat, Aeneas, or Gentle, this should be straightforward. Let me know your preferred toolchain, a rough turnaround per recorded hour, and a brief note on past projects that demanded word-level accuracy.
transcription json audio processing audio editing audio engineering speech synthesis data annotation speech recognition
Apply on Freelancer →

Project sourced from Freelancer.com. Applications happen directly on the original platform — we never collect your data.