Everything you record, in writing.

Pilcrow transcribes your recordings, then does the parts other tools leave to you: from insights tailored to each recording to an assistant that analyses and acts across your whole library.

Insights that adapt to each recording.

Pilcrow reads your transcript and decides what to surface: decisions and action items where things were agreed, unanswered questions where they weren't, a clean write-up when you were dictating. Add your own insights with a one-line description.

Dictation · Jun 22, 2026
Palace Cinema — structural survey
26:47 1 English

Auditorium roof first. Active water ingress over the balcony — the parapet gutter on the north slope has failed, and the boarding under the felt gives along a two-metre run.

Front elevation now. One diagonal crack in the render above the entrance canopy, hairline, tracking the steel behind. No movement in the brickwork, so I'm minded to read it as thermal. Flag for monitoring.

Into the foyer. The cornice plaster has blown in two places by the kiosk, and there's damp staining at the foot of the stair wall — localised, and consistent with the failed downpipe outside.

Up into the roof void to finish. The trusses read sound and the boarding is dry away from the parapet, but there's daylight showing at two of the slate verges on the north side.

A full walkthrough of the former Palace Cinema, dictated on site. The structure is broadly sound, but the parapet gutter over the balcony has failed and water is reaching the roof timbers — the one item to address before winter.

The remaining defects are weathering or cosmetic: a hairline render crack over the entrance canopy, corroded steel window frames to the first-floor offices, and blown plaster in the foyer.

The projection room was locked on the day of inspection; a follow-up visit is recommended once access is arranged.

Six defects were recorded, in order of severity:

  • High — failed parapet gutter over the balcony 04:12, with water ingress softening the roof boarding.
  • Medium — diagonal render crack above the entrance canopy, hairline, likely thermal 06:48.
  • Medium — corroded steel window frames to the first-floor offices, with two cracked panes 11:20.
  • Low — blown cornice plaster in the foyer, by the kiosk.
  • Low — damp staining at the foot of the stair wall.
  • Low — decayed timber to the west fire-exit doors 19:35.
Element Measurement Notes Timestamp
Ridge height 15.8 m Over the auditorium 02:36
Floor area 940 m² Stalls and balcony 09:14
Render crack 1.2 m Hairline, above the canopy 06:48
Moisture 31% WME Roof boarding at the parapet 04:41
External walls 330 mm Solid brick, rendered front 08:03
Softened boarding 2.0 m Along the parapet gutter 04:12
Damp staining 0.6 m Foot of the stair wall 13:05
  • Renew the balcony parapet gutter and replace the softened boarding before winter.
  • Rake out and monitor the render crack; review in six months.
  • Re-fix the blown cornice plaster at the next redecoration.
  • Commission a measured drone survey of the full roof.
  • Repair the failed downpipe and repeat the moisture readings after heavy rain.
  • Overhaul the corroded office window frames and replace the cracked panes.
  • Arrange access to the projection room and complete the inspection.

An assistant that acts, not just answers.

Ask a question and you get an answer, with the excerpt and timestamp to back it. Give an instruction and it gets done: search, edit, delete, file, tag, translate, export, and more. On a transcript, project, or your entire workspace.

Workspace assistant
Pull the Meridian calls into a project, tag each one by meeting type, and pin the ones where pricing came up.
Today 1/3
Searched the library · 6 matches
Created project Meridian
Moved 6 recordings
Tagged 6 recordings
Pinned 2 recordings

Done — six recordings are filed under a new Meridian project and tagged by meeting type. Pricing was mentioned in two of them, so both are pinned to the top: Kickoff and Pilot review.

For last week’s user interviews, pull out each feature request and who asked for it. Do this for every user interview going forward.
Today 2/3
Searched the library · 4 matches
Read 4 recordings
Created 4 insights
Saved a standing instruction

I created four insights, one on each interview, listing its feature requests and who made them. The standing instruction is saved, so every new interview gets one automatically.

Speaker 2 in this morning’s stand-up is Priya, and the intro credited to Marcus was actually her. Fix both, change every “Meridien” to “Meridian”, and export the transcript as a PDF.
Today 3/3
Renamed Speaker 2 to Priya
Reassigned 3 segments
Edited 6 segments
Exported transcript · PDF

Done — Priya is named on all 14 of her turns, including the three intro segments credited to Marcus, and all six “Meridien”s now read “Meridian”. The corrected transcript is in your downloads.

Ask across your recordings, or tell me what to do…
Standard

Projects that give your recordings context.

A project's instructions, background, and vocabulary follow its recordings through everything Pilcrow does with them: transcription, titles, tags, insights, translations, and the assistant's answers. Write them once; every recording filed there arrives with that context already applied.

Everything a transcription tool should do.

Capture, speaker-labelled transcripts, editing, search, translation, export, sharing, and more.

Upload files or record in the browser

Upload audio or video, or record your microphone, a browser tab, or your entire screen.

Speaker-labelled transcripts

Each voice is detected and given its own colour, with names or roles inferred from the conversation itself.

Transcripts synced to the audio

The active word highlights as the audio plays. Click any word to jump to the moment it was said.

Edit the transcript in place

Fix anything the transcription got wrong. Word timestamps are kept through the edit, so synced playback and subtitles still line up.

Automatic titles, tags, and descriptions

Every recording arrives with a title, description, and tags already written. No untitled uploads to sort out later.

Search everything you've captured

Everything you record builds into one searchable knowledge base. Describe what you're looking for in your own words, and the matching moments come back.

Translation into 50 languages

Read any transcript side by side with a translation, speaker for speaker.

Export in 8 formats

Take any transcript or insight out as TXT, PDF, DOCX, Markdown, HTML or JSON, or as SRT and VTT subtitles.

Share any recording with a link

When published, anyone with the link can read the transcript, insights, and translations, and play the synced audio, no account needed.

The whole product on every plan.

Start free with 2 hours of transcription a month. Upgrading gets you more usage, faster processing, and priority support.

Free Plus Custom
Price $0 forever$14 / monthContact for pricing
All features IncludedIncludedIncluded
Transcription 2 hrs / month20 hrs / monthUnlimited
Assistant questions 20 / day100 / dayUnlimited
Translations 1 per transcript3 per transcriptUnlimited
Processing StandardPriorityDedicated
Support StandardPriority emailDedicated
Bespoke features & integrations Included

Free

$0 forever
All features
Included
Transcription
2 hrs / month
Assistant questions
20 / day
Translations
1 per transcript
Processing
Standard
Support
Standard
Bespoke features & integrations

Plus

$14 / month
All features
Included
Transcription
20 hrs / month
Assistant questions
100 / day
Translations
3 per transcript
Processing
Priority
Support
Priority email
Bespoke features & integrations

Custom

Contact for pricing
All features
Included
Transcription
Unlimited
Assistant questions
Unlimited
Translations
Unlimited
Processing
Dedicated
Support
Dedicated
Bespoke features & integrations
Included

Common questions.

Can't find your answer? Contact us.

What can I upload?

Audio or video files up to 1GB and 3 hours long. You can also record directly in the browser: your microphone, the audio of a browser tab, or your entire screen.

Is my audio used to train AI models?

No.

How is my data protected?

Recordings and transcripts are encrypted in transit and at rest, audio is served through signed, time-limited URLs, passwords are stored hashed, and access to your content is scoped to your account.

How accurate are the transcripts?

Pilcrow transcribes with a state-of-the-art speech model at roughly 95% word accuracy on clear English audio, with comparable results across other supported languages.

Which languages can be transcribed?

Pilcrow detects the spoken language automatically and transcribes the recording in it. There's nothing to configure. Titles, descriptions, and insights are written in the transcript's language too.

Can I cancel Plus at any time?

Yes. Cancel from your billing settings whenever you like. Plus stays active until the end of the period you've already paid for, and after that your account moves to the Free plan.

What happens if I downgrade from Plus to Free?

Nothing is deleted or locked: every transcript, translation, and insight you've created stays fully available. Once the paid period ends, the Free plan's limits apply to new transcription hours, assistant questions, and translations.

Can I get my data out?

Always. Export transcripts or insights in up to 8 different formats, or download everything in your account as a single archive.

Get on the waitlist.

Pilcrow is in closed beta. Leave your email and we'll send an invite as spots open.

You're on the list. Watch your inbox for an invite.