Here’s what I turn your speech into.

Subtitle the film. Transcribe the interview. Capture the meeting. Caption the room.

I subtitle your film to spec, in every language you ship

Dual ASR engines: Two engines run per language pair, with millisecond timecodes and silence recovery.
Glossary enforced: Every term corrected before translation — each substitution kept on the record.
Export to spec: SRT, VTT, FCPXML, XLSX, Markdown, plus a burned-in master up to 4K.

Start for free See how it works

Subtitle this film with the Enhanced model, English into Español, keep the Launch-24 glossary on.

MP4

launch-film.mp4

42:181.9 GB4K

Transcribedual ASR · millisecond timecodes · silence recovery

Apply glossary64 terms enforced · every fix logged

Export to specSRT · VTT · burned-in up to 4K

Subtitling 47 cues⚡ Proofread 96/100✓ Glossary applied

#TimecodeCue · EN → ESStatus

04400:10:55,120We spent two years on one question.Pasamos dos años en una sola pregunta.Verbatim

04500:10:58,760How does a film ship in every language at once?¿Cómo se estrena un filme en todos los idiomas a la vez?Timed

04600:11:02,340Welcome back — Datax ships to every studio today.Bienvenidos — Datax llega hoy a todos los estudios.Fixed

04700:11:05,880The new pipeline cuts a two-week edit to an afternoon.El nuevo flujo reduce dos semanas de edición a una tarde.Timed

04800:11:09,120Every caption stays in sync with the master.Cada subtítulo permanece sincronizado con el máster.Timed

04900:11:12,600The hardest part was earning the first ten clients.Lo más difícil fue ganar los primeros diez clientes.Verbatim

05000:11:16,040So we hand you a file you can actually publish.Así que entregamos un archivo que realmente puedes publicar.Verbatim

05100:11:19,480Spelled to spec, timed to the frame.Con la ortografía exacta y el tiempo al fotograma.Timed

05200:11:22,900That is the whole promise.Esa es toda la promesa.Verbatim

I keep every speaker straight and every word quotable

Word-level diarization: Overlapping voices split at word boundaries, each cue owned by one speaker.
Named speakers: Roster names applied across every cue and carried through every export.
Delivered by the afternoon: DOCX, XLSX, TXT, and a ZIP with both languages side by side.

Start for free See how it works

Transcribe this morning’s interviews with 2 speakers, verbatim — deliver DOCX and a bilingual ZIP.

M4A

interview-02.m4a

2:12:04640 MB48 kHz

Diarizeword-level speaker separation

Labelroster names applied

DeliverDOCX · XLSX · ZIP

Labeling 2 speakers⚡ 24 turns✓ Quotable

SpeakerVerbatimTimecode

May HoThanks for making the time — I know the round has kept you busy.00:04:12

Daniel ReyesOf course. Northwind's been on my radar since the pilot in Leeds.00:04:19

May HoThat pilot is exactly why the timeline matters now.00:04:27

Daniel ReyesSo walk me through what the raise actually unlocks for you.00:04:34

May HoWe ship to three cities by spring — the funding just makes the timeline honest.00:04:41

Daniel ReyesHonest is the word every founder skips. What breaks if you slip a quarter?00:04:52

May HoNothing breaks — but the newsrooms who bet on us start hedging.00:05:01

Daniel ReyesFair. And the hires are editorial-first, not just engineering?00:05:10

May HoEditorial-first, always. The product is only as good as the desk running it.00:05:18

I capture your meetings, with a bot or without one

Notetaker bot, calendar aware: Joins Meet, Zoom, or Teams five minutes early — and keeps the record even when you can’t attend.
No-bot extension: The Chrome extension captures your own tab when nothing should appear in the participant list.
Decisions with owners: Action items, chapters, and a record the whole team can search.

Start for free See how it works

Capture my 9:30 standup — no bot joining — and send decisions with owners to the team by lunch.

MEET

meet.google.com/kzo-qhma-fyx

9:30 standup5 participants32 min

CaptureChrome extension · tab audio + roster

Diarizereal names from participants

Summarizedecisions · action items · chapters

Summarizing 32 min✓ 2 decisions⚡ 3 action items

Decisions

Ship the beta on March 14.Sent

Localized captions move to v1.1.Sent

Action items

Send the revised pricing deckPNPriya N.Fri

Draft release notesMRMarco R.Mon

Book the QA pass for captionsDWDana W.Review

The room speaks, and I write it down as it happens

Speech is the request: No prompt, no upload — each spoken line drafts, then inks into a timecoded record.
Translated as you talk: Español runs live beside the original, with more languages queued behind it.
The archive builds itself: Every settled line lands on the record — by the time you sit down, the transcript is done.

Start for free See how it works

Live · the room is talking

Shure MV7 · Main hall

Caption214 settled · millisecond timecodes

TranslateEspañol live · +3 queued

Archivebuilds as you speak

Session 42 min⚡ 5 languages✓ Ready to export

#TimecodeLine · EN → ESStatus

04200:42:11,000Good morning, and welcome to the Product Summit.Buenos días y bienvenidos a la Cumbre de Producto.Settled

04300:42:16,480Today we ship to every studio, in every language at once.Hoy lanzamos a cada estudio, en cada idioma a la vez.Settled

04400:42:22,120The room reads along on their own screens as I talk.La sala lee en sus propias pantallas mientras hablo.Settled

04500:42:29,640This is where Datax earns its place on the record.Aquí es donde Datax se gana su lugar en el registro.Live

04600:42:35,300Every phrase is timed to the millisecond as it lands.Cada frase queda cronometrada al milisegundo al llegar.Settled

04700:42:41,020Español runs live; three more queue behind it.El español va en vivo; otros tres esperan detrás.Settled

04800:42:47,560The archive builds itself while I keep speaking.El archivo se construye solo mientras sigo hablando.Settled

04900:42:53,900By the time I sit down, the transcript is done.Para cuando me siente, la transcripción estará lista.Settled

95+ languages, one standard.

However your audience speaks, the deliverable holds the same bar.

98%

average word accuracy — measured across everything we transcribe, code-switch included

廣東話
English
普通话
國語
Español
Français
Deutsch
日本語
한국어
Português
Italiano
Nederlands
Svenska
廣東話
English
普通话
國語
Español
Français
Deutsch
日本語
한국어
Português
Italiano
Nederlands
Svenska

हिन्दी
Tiếng Việt
ไทย
Bahasa Indonesia
Bahasa Melayu
Filipino
العربية
עברית
Türkçe
Ελληνικά
Українська
Polski
हिन्दी
Tiếng Việt
ไทย
Bahasa Indonesia
Bahasa Melayu
Filipino
العربية
עברית
Türkçe
Ελληνικά
Українська
Polski

Čeština
Română
Magyar
Dansk
Norsk
Suomi
বাংলা
اردو
ខ្មែរ
Монгол
Kiswahili
Español · LatAm
Čeština
Română
Magyar
Dansk
Norsk
Suomi
বাংলা
اردو
ខ្មែរ
Монгол
Kiswahili
Español · LatAm

Translation that reads the whole story.

Line-by-line tools lose track of who and what — “she” becomes “he” by line two. Subanana carries the whole transcript.

Original text

Mi hermana terminó el documental en marzo.

Volvió agotada del rodaje.

El equipo pasó meses en la montaña.

Aún no lo he visto entero.

Se estrena el mes que viene.

Espero que llene salas.

Gemini

My sister finished the documentary in March.

He came back exhausted from the shoot.

The crew spent months in the mountains.

I haven’t seen him in full yet.

It premieres next month.

I hope it fills theaters.

ChatGPT

My sister finished the documentary in March.

He came back exhausted from the shoot.

The crew spent months in the mountains.

I haven’t seen it in full yet.

It premieres next month.

I hope it fills theaters.

My sister finished the documentary in March.

She came back exhausted from the shoot.

The crew spent months in the mountains.

I haven’t seen it in full yet.

It premieres next month.

I hope it fills theaters.

One platform. Every spoken-word deliverable.

Pick a job. Hand it to Subanana.

Cue agentsclean every line

From raw audio to publishable cues in every export format.

Subtitlethis launch videoin95+ languages

withglossary terms lockedthenexport to spec

🇪🇸ES🇺🇸ENMP4🇫🇷FR🇯🇵JA🇭🇰YUE

Burned in

Welcome to our spring launchBienvenidos a nuestro lanzamiento

Caption style

37 presets

Start from a preset or set it yourself — font, outline, shadow, plate, glow and position. Each language keeps its own style, and yours saves as a workspace template.

AI correction

34 fixes found. In the editor, suggestions apply only when you approve them.

Name spelling

north windNorthwind

6

Filler words

we, uh, rebuiltwe rebuilt

14

Punctuation

is it live yetis it live yet?

9

Grammar

runs more fasterruns faster

5

Export to spec

Every deliverable from one track, timecoded to the millisecond.

SRT

SubRip

VTT

WebVTT

TXT

Plain text

DOCX

Word

XLSX

Excel

MD

Markdown

MP4

Burned-in MP4

Bilingual, styled to your spec, up to 4K

SRT

northwind-spring-update.en-es.srt

1

00:00:12,480 --> 00:00:15,120

Welcome to the Northwind spring update.

Bienvenidos a la actualización de primavera.

2

00:00:15,400 --> 00:00:18,060

We rebuilt the editor from the ground up.

Reconstruimos el editor desde cero.

3

00:00:18,200 --> 00:00:21,640

Every file you export now matches your spec.

Cada archivo cumple tu especificación.

Translationthat stayson-brand

Turn one transcript into

95+ languages.

38 / 40 Locked

95%glossary hit

Check every term against your glossary.

Sarah Chan

🇺🇸 EN → 🇭🇰 ZH

Glossary OK

Marcus Lee

🇺🇸 EN → 🇯🇵 JA

Glossary OK

Wong Ka-yan

🇺🇸 EN → 🇪🇸 ES

2 terms open

Localize

Glossary terms, register, formats, and beyond.

Glossary Terms

Locked: Datax, Subanana, Kowloon

Language Pair

English → Traditional Chinese (HK)

Formats

SRT, VTT, DOCX, burned-in

Register

Formal tone, honorifics kept

Proofread

Every segment gets an AI proofread score.

90-100 · Clean

18

75-89 · Minor fixes

9

50-74 · Needs a pass

8

Below 50 · Reword

12

Share

Share each track with a teammate for a check.

Auto-approve clean passes18

Share with teammate9

Flag for a second look20

Teammates on this file

Chan Mei-lingCantonese6

Kenji SatoJapanese6

Ana CruzSpanish6

Deliver

Send each localized file to spec.

To

editor@subanana.com

Cantonese subtitles · SRT + burned-in MP4

Hi editor,

Your Cantonese pass is glossary-clean.

Glossary terms are locked, speaker labels carried across, and the SRT cue timings match the source exactly.

Want the Japanese and Spanish passes next?

Best,
Nadia

AI agentsresearch yourinterviews

Know who said what,

and exactly when.

Surface quotable segments

Transcript cues

382

47:12of audio

Speakers separated

4

Guest, Host+ S1, S2

Search matches

12

“insight”in 3 speakers

Interview 07 segment summary

Speaker 173

Speaker 257

Guest147

Host105

Quoted segments

“We run three interviews a week and never type them up”00:03:18

“Turnaround under an hour changed our workflow”00:21:47

Speaker Snapshot

S1

Speaker 1 · Dr. Chan

Interview 07 · Media policy

Key Timecodes

00:07:14 “we need it word for word”

00:24:52 “play it back at the timecode”

Topics

Language depthTurnaround

SpellingGlossarySpeaker labels

Segments

Speaking time

How the 47 minutes split across 4 separated speakers

47

min

Minutes per speaker

Guest18

Host13

Speaker 19

Speaker 27

Quote finder

Search a term, get verbatim cues with speaker and timecode

MatchVerbatimTimecodeCues

“budget”

“We just don’t have the budget for a full desk.”

00:04:12.480 · Speaker 2 · verbatim cue

3 cues

“Cantonese”

“It has to handle Cantonese, not just Mandarin.”

00:09:31.120 · Guest · speaker-labeled cue

2 cues

“hour”

“We need a clean draft within the hour.”

00:12:40.960 · Host

timecoded to the millisecond

2 cues

“jargon”

“The names and jargon must be spelled right.”

00:18:05.240 · Speaker 1 · glossary-enforced

1 cue

AI agentsrecord everymeeting

Bot in the call, or

no bot at all.

Meeting notes for the Q3 sync

Platform

Meet · Notetaker

Attendees

6 joined4 spoke

Recorded

Jul 23, 2026

Duration

32 minutes

Meeting Snapshot

Weekly standupProduct team

Google Meet · 09:30recorded

3Decisions

Action items

+3

Decisions

Ship beta on Friday

Freeze scope this sprint

Who Joined

Sarah Chen

Product lead · meeting host

MeetTeamsZoom

Transcript

Sarah: let’s ship the beta Friday

2d ago

Marcus: I’ll own the QA pass

2d ago

What the Meeting Decided

Decisions Captured

Ship the beta on Friday

QA pass owned by Marcus, due Thursday

Freeze scope this sprint

No new tickets until the beta ships

Move the demo to Wednesday

Marcus sends the invite after standup

Action Items & Responses

Draft the release notes

Owner: Priya · due Friday · pulled from the transcript.

In progress

Book the demo slot

Owner: Marcus · due Wednesday · flagged in the notes.

Not started

Searchable Transcript

Search “beta date”

Jump to a decision

Find every action item

Q3 sync - Meeting notes

DOCX

Open

AI agentscaption liveevents

Pre-load speaker names and jargon, then caption live to every screen in the room.

Captioning Live to Every Screen in the Room

Captionthis keynotetoscreens + attendee phones

withglossary pre-loadedinEnglish + Cantonese

Latency < 2.0s12 languages

On the Event Console

Alexandra Lee

Keynote speaker · pre-loaded

DataxCantonese jargonCantonese

Words8.2K

Glossary128 terms

Latency1.4s

Live Caption Preview

Live

Welcome to the spring
developer keynote, everyone

00:04:12 · shown on screen

Live

First up, what changed
in the API this spring

00:04:19 · shown on screen

Speakers Loaded for This Session

SpeakerSetupWordsTermsLatencyTrack

Alexandra Lee

Ready42.7K1281.4sKeynote

James Carter

Ready37.2K961.6sPanel

Maya Patel

Ready28.6K741.8sBreakout

Tyler Nguyen

Ready31.4K611.9sWorkshop

Sara Lam

Prep26.1K482.1sFireside

Share Captions to Every Phone

Scan the QR code to read captions on your phone.

James Carter

Scan to join

Captions appear on your phone within two seconds, with speaker names and jargon already loaded for this session.

Read along in English or Cantonese, switch languages any time, and take home a speaker-labeled transcript afterward.

Alexandra Lee

Scan to join

Point your camera at the code on screen, or tap the link the host shares in the room.

Captions stream live to your phone in English or Cantonese, with speaker names and jargon already loaded.

Prefer a transcript afterward? It exports to SRT, VTT, or DOCX.

AI agentsexport toany format

Every deliverable, ready

in the format you hand off.

Exporting every deliverable to spec

Formats Ready

7

SRT, VTT, and more

Timecodes

412

To the millisecond

Speaker Labels

4

Through every export

Formats selected

Which specs this export set covers

Your file

Keynote · English + Cantonese

Export set

SRT · VTT · DOCX · MP4

4 / 7

Top export preset

B

Broadcast SRT

Broadcast caption spec

SRT25 fpssidecar

Format spec

SRT

Timecode

±1ms

Frame rate

25 fps

Why this preset passes QC

Timecodes land to the millisecond
Speaker labels survive the export
Matches the broadcast caption spec

Best presets to export with

PresetFormatTimecodeBest forStatus

BBroadcast SRT

SRTSubRip

±1 ms

Broadcast

Ready

WWeb VTT

VTTWebVTT

±1 ms

Web player

Ready

VVerbatim DOCX

DOCXWord

Per cue

Records

Ready

BBurned-in MP4

MP4H.264

Burned in

Social cut

Ready

BBilingual SRT

SRTSubRip

±1 ms

Two languages

Ready

Delivering the finished files

Links, not attachments

To

producer@subanana.com

Keynote exports ready · 412 cues

Hi producer,

Your keynote is transcribed and glossary-checked — 412 cues across English and Cantonese, timecoded to the millisecond.

Every format is ready to download: SRT, VTT, TXT, DOCX, XLSX, MD, and a burned-in MP4. Speaker labels carry through all of them.

The links open straight into your workspace, so the team can pull the same files without waiting on an attachment.

One platform. Every spoken-word deliverable.

Pick a job. Hand it to Subanana.

Cue agentsclean every line

From raw audio to publishable cues in every export format.

Subtitlethis launch videoin95+ languages

withglossary terms lockedthenexport to spec

🇪🇸ES🇺🇸EN🇫🇷FRMP4🇭🇰YUE

Burned in

Sarah Chan

Host · English

🇺🇸 EN → 🇪🇸 ES412 cues

37 presetsPer languageTemplates

Cue 001 · 00:00:12 → 00:00:15 · 🇺🇸 EN

“Welcome to our spring launch”

🇪🇸 ES keeps its own style

AI correction

34 fixes found. In the editor, suggestions apply only when you approve them.

Name spelling

north wind → Northwind

6

Glossary term

Filler words

we, uh, rebuilt → we rebuilt

14

Cleaned up

Punctuation

is it live yet → is it live yet?

9

Repaired

Grammar

runs more faster → runs faster

5

Revised

Export to spec

SRTVTTMP4TXTDOCXXLSXMD

SRT

northwind-spring-update.en-es.srt

Launch video · 412 cues · 🇺🇸 EN → 🇪🇸 ES

1

00:00:12,480 --> 00:00:15,120

Welcome to the Northwind spring update.

Bienvenidos a la actualización de primavera.

2

00:00:15,400 --> 00:00:18,060

We rebuilt the editor from the ground up.

Reconstruimos el editor desde cero.

The last mile between speech and publishable

Built for hard languages. Judged at the deliverable.

The engine underneath.

Connect anything. Edit in one place.

It tells you what it changed.

Accuracy Lift

Cantonese

98.2%

Speaker labeled

Enforced

98.2%

4 minutes ago

New glossary term enforced

1 hour ago

Exported a burned-in MP4

to spec, on deadline

Cantonese glossary locked last week. Cue accuracy up 18% after speaker labels applied.

4 minutes ago

Languages served

95+

4 minutes ago

Star Ferry crossing124 cuesSaved

Timing🇬🇧 English · source🇭🇰 繁體中文CPS

00:12.400:15.1The Star Ferry has crossed this harbour since 1888.天星小輪自 1888 年起橫渡這個海港。14

00:19.000:22.6It still costs less than a bottle of water.票價仍然比一支水便宜。17

00:27.200:29.9Sit on the upper deck if you want the view.想看風景就坐上層。13

00:34.000:36.8Crossings run every ten minutes or so.航班大約每十分鐘一班。12

00:41.500:44.9The whole trip takes about eight minutes.全程大約八分鐘。11

00:48.200:51.0Go at dusk, when the lights come on.黃昏時去，華燈初上。15

00:55.600:58.4It is the cheapest harbour tour in the world.這是全世界最便宜的海港遊。13

01

Connect anything. Edit in one place.

02

It tells you what it changed.

03

The engine underneath.

Accuracy Lift

Cantonese

98.2%

Speaker labeled

Enforced

98.2%

4 minutes ago

New glossary term enforced

1 hour ago

Exported a burned-in MP4

to spec, on deadline

Cantonese glossary locked last week. Cue accuracy up 18% after speaker labels applied.

4 minutes ago

Languages served

95+

4 minutes ago

Star Ferry crossing124 cuesSaved

Timing🇬🇧 English · source🇭🇰 繁體中文CPS

00:12.400:15.1The Star Ferry has crossed this harbour since 1888.天星小輪自 1888 年起橫渡這個海港。14

00:19.000:22.6It still costs less than a bottle of water.票價仍然比一支水便宜。17

00:27.200:29.9Sit on the upper deck if you want the view.想看風景就坐上層。13

00:34.000:36.8Crossings run every ten minutes or so.航班大約每十分鐘一班。12

00:41.500:44.9The whole trip takes about eight minutes.全程大約八分鐘。11

00:48.200:51.0Go at dusk, when the lights come on.黃昏時去，華燈初上。15

00:55.600:58.4It is the cheapest harbour tour in the world.這是全世界最便宜的海港遊。13

Connect anything. Edit in one place.

A file, a YouTube or Instagram link, a Meet, Zoom or Teams call, OBS, or the browser recorder. Whatever the source, it lands in one editor as cues you can correct.

It tells you what it changed.

Glossary terms enforced, speakers labelled, accuracy moved — every pass reports what it did, so approving it is a decision rather than an act of faith.

The engine underneath.

95+ languages to a publishable standard, code-switched Cantonese included, timecoded to the millisecond — and exports that survive QC in every format you hand off.

Don’t take our word for it.

Ask an assistant what we do. These are questions with answers you can check for yourself.

What is Subanana used for?

Ask it in

ChatGPT Claude Perplexity

ChatGPT, Claude and Perplexity are trademarks of their respective owners and are not affiliated with Subanana.

Spoken content, ready for every audience.

Subtitles & Translation

Meetings & Notetaker

Live Captions & Events

Interview & Research

Trusted where accuracy is the deliverable.

Here’s what I turn your speech into.

I subtitle your film to spec, in every language you ship

95+ languages, one standard.

Translation that reads the whole story.

One platform. Every spoken-word deliverable.

Cue agentsclean every line

Burned in

AI correction

Export to spec

One platform. Every spoken-word deliverable.

Cue agentsclean every line

Burned in

AI correction

Export to spec

The last mile between speech and publishable

Connect anything. Edit in one place.

It tells you what it changed.

The engine underneath.

Don’t take our word for it.

From speech topublishable.

Spoken content,
ready for every audience.