ingest · pdf and audio

PDF and audio in. Exact, anchored data out.

Orves Ingest is an async API that starts with PDF, born-digital and scanned, and with audio and podcasts. Its output is designed to be deterministic, with an exact anchor on every element back to where it came from.

Get a demo Talk to us

PDF · born-digital + scannedaudio and podcastsdesigned for deterministic outputdesigned for exact anchorsasync

illustrative · not a final schema

# a PDF element and its exact anchor
{ "text": "Renewal date: September 1, 2026",
  "anchor": { "page": 4, "box": [72, 310, 410, 332] } }

# an audio segment and its time anchor
{ "text": "…renewal may be postponed again",
  "anchor": { "start": "00:14:32", "end": "00:14:36" } }

What it produces

Structure you can point back to.

PDF, born-digital and scanned

text
tables
reading order

Digital and scanned documents are designed to come back as structured, ordered content.

Audio and podcasts

transcript
segments
time anchors

Spoken content is designed to become text that points to the moment it was said.

Designed to be deterministic and anchored

same input
same output
exact anchors

By design, the same input gives the same output, and every element anchors to its place in the source.

How it works

From input to anchored output.

PDF or audio in→Orves Ingest→async job→anchored output→Orves Kore

Benchmarks

Measured before claimed.

We publish a quantitative claim only when a benchmark receipt exists and is linked. None exists yet, so no number appears here. Benchmark coming.

How we benchmark →

API first

Submit, poll, fetch.

Async by design. Planned delivery is through the managed API, with nothing to host.

REST APIasync jobsbatch jobs