Recording
Your recording
Paste your recording. Nothing to upload.
←recording
Make a financial analysis
Turn a recording into a financial analysis
A number spoken on a call and a number shown on the slide are not always the same number.
Free to try, no card. Your recording opens in Ancher.
Transcript with timingsCommitment languageUnresolved threads
Kept in parts
Every one of these stays separately addressable instead of collapsing into one block of text.
Headline figuresEvidence per figureTrendRisks
Financial analysis
Figures checked against the sentence or table that supports them.
Financial figures said out loud are rounded, hedged and sometimes restated three minutes later. An analysis built from audio alone inherits every one of those distortions silently. Ancher keeps each figure attached to the second it was spoken, keeps the frame if it was also displayed, and flags where the spoken version and the on-screen version disagree instead of quietly picking one.
The honest version
When the spoken figure and the slide disagree, both are kept
On any call with slides, someone says roughly a third while the chart behind them reads 31.4%. Neither is wrong, but only one belongs in an analysis, and a transcript-only tool never sees the second. Ancher holds both, marks the divergence, and leaves the judgement to you rather than resolving it out of sight.
One second of one recording, at 08:14, read two ways.Illustrative. The figure is invented; the gap it falls into is not.
gross margin was roughly , call it flat year on year
Slide displayed at the time — not readThe specific never reaches the draft
Financial analysis
Slide displayed at the time
GM 62.4% (prior year 64.1%)
from 08:14
Lands in the draft, with its timecode
- Every figure carries the second it was said, so a rounded number can be re-heard in its original sentence.
- Where a figure was also displayed, the frame is kept alongside the spoken version.
- Hedges survive into the analysis — roughly, about, on a normalised basis — because dropping them is how a caveat becomes a claim.
What comes out
Financial analysis, section by section
Figures checked against the sentence or table that supports them.
Headline figures
What the numbers amount to, in one paragraph, before any table.
Evidence per figure
Each figure with the sentence it came from and the second it was said.
Trend
Direction over the periods discussed, with what was and was not comparable.
Risks
Caveats as stated, including anything the speaker declined to answer.
Sources
Which recording, what date, and where a figure was spoken versus shown.
Recording-to-financial-analysis prompt
The link on its own gets a summary. This is the instruction that produces the 5 sections above, in that order, with the rules that keep them honest. Paste it with your recording — in Ancher, or in whatever assistant you already use.
Analyse the figures in this recording. Open with what they amount to in one paragraph. List every figure with the sentence it came from and its timestamp, and mark whether it was spoken, displayed, or both. Where the spoken and displayed versions differ, keep both. Preserve hedges exactly as stated, note anything the speaker declined to answer, and identify what was not comparable across periods.
The detail, if you want it
How each part of a recording becomes a section
| From the recording | Into | Why |
|---|---|---|
| Transcript with timings | Evidence per figure | A figure without the sentence around it loses its basis, and the timing is what keeps the sentence recoverable. |
| Screen shares, if it is video | Evidence per figure | Displayed figures are the precise ones; spoken figures are the rounded ones, and an analysis needs to know which it is holding. |
| Unresolved threads | Risks | A question asked on a call and not answered is itself a finding, and it never appears in a summary. |
| Commitment language | Trend | Guidance is commitment language — we expect, we will be at — and it is what separates a forecast from a comment. |
| Duration and timeline | Headline figures | The line item that took the most airtime is usually the one under pressure, which is a reasonable place to start the headline. |
Everything Ancher keeps from a recording
Transcript with timings
Every segment carries start and end seconds, which is what turns a claim into a record.
Commitment language
I'll take that, by Friday, let's park it — these phrases are what separate a task list from a topic list.
Unresolved threads
What a conversation failed to close is usually the reason for the next one.
Screen shares, if it is video
A shared screen is sampled as frames, so the number that drove a decision stays attached to it.
Duration and timeline
Where the time actually went is a finding in itself for any recurring meeting.
The four steps
- Add the recording. Upload the audio or video file. Ancher transcribes it and keeps a start and end time on every segment, so any line in the output can be traced back to the moment it was said.
- Nothing gets flattened. Transcript with timings, commitment language, unresolved threads, screen shares, if it is video, duration and timeline stay separately addressable instead of collapsing into one block of text.
- Ask for a financial analysis. Ancher fills headline figures, evidence per figure, trend, risks, sources using the mapping above, not a generic summary pass.
- Check it before you use it. Every specific keeps a link back to where it came from in the recording, so a claim can be verified without reopening the source.
Where this goes wrong
- Spoken figures are approximations. Anything used for a decision should be checked at its timestamp, and ideally against a filed document.
- Without speaker labels, a figure cannot be attributed to a named executive from the audio alone. Attribution needs your own listen.
- This is a reading of what was said on a recording. It is not investment advice, and it is not a substitute for audited statements.
Questions people ask
What happens when the number changes mid-call?
Both versions are kept with their timestamps and the later restatement is marked as such. Speakers correct themselves constantly on live calls, and an analysis that silently keeps the first figure is wrong in a way nobody can see.
Can it read numbers off the slides?
If the recording is video, frames are sampled, so a figure that was displayed but never read aloud is available. On audio-only recordings you get only what was spoken, and the analysis says so.
Is this good enough to trade on?
No. Treat it as a fast structured read of what was said, with every claim traceable to a second of audio, and verify anything material against the filings before it informs a decision.
Bring your own recordings
Everything you save lives in one workspace, so the financial analysis is built from your sources — not from a model's memory of the internet.
Open Ancher →Not a recording? The same financial analysis also comes from YouTube video, X thread, PDF.