How do I find where a specific word was said across many recordings?

Transcribe the library once with an on-device tool, then search the index. MediaFind returns the exact timestamp in the exact file for a word, phrase or paraphrase, across your entire archive, without uploading anything.

Can I search by meaning instead of the exact words?

Yes — semantic search finds moments by meaning ('the part about pricing objections') while exact search finds literal words and names. MediaFind runs both and fuses the results.

Can I filter spoken-word search by who said it?

Yes. Speaker identification labels who is talking, so you can search for what a specific person said — 'moments where Sarah mentions the deadline' — across every recording.

Find any spoken word across hours of video or audio

The quote exists. It's in one of forty interviews, or a hundred episodes, or three years of lecture recordings — and scrubbing through footage at 2× is not a retrieval strategy. This is the actual workflow for finding the moment anything was said.

Step 1 — transcribe everything, once

On-device Whisper-class models make this practical now: roughly real-time per file on an ordinary machine, no per-minute API fees, nothing uploaded. MediaFind does this automatically for every file you point it at, and picks up new recordings incrementally.

Step 2 — search words and meaning

Exact search finds literal words and names ("Meridian", "clause 7"). Semantic search finds the moment by meaning — "the part where he explains why the deal died" — even when you can't remember the phrasing. MediaFind runs both over the whole library and fuses the results, so a plain-language query returns ranked moments with timestamps, not a pile of documents.

Step 3 — filter by who said it

Speaker identification labels voices across your library, so "moments where the guest mentions funding" and "everything Sarah said about the deadline" are queries, not an afternoon of listening.

Step 4 — use the moment

Click a result to play from the exact second. Export the clip, or hand the moment to Final Cut, Premiere or Resolve. For podcasters that's pull-quotes and promo clips; for journalists, the verifiable quote with its source; for lecturers and researchers, the citation with a timestamp.

Who this matters for: podcasters mining a back catalog, journalists verifying quotes, lawyers reviewing depositions, lecturers reusing past teaching, support and sales teams mining calls — anyone whose archive is full of words they can't currently find. All of it stays on your machine: no upload, no account, and a built-in audit that proves it.

Find any spoken word across hours of video

Step 1 — transcribe everything, once

Step 2 — search words and meaning

Step 3 — filter by who said it

Step 4 — use the moment

Search your whole library for $29 — once.