Opportunities › Speaker identification for meeting transcription › Proposed app
Who Said
Private meeting transcripts that keep every speaker straight, no calendar needed
- Score
- 43
- confidence 46
- Evidence
- 7
- items on the page
- Rivals
- 1
- named in sources
- First version
- 40 d
- estimate
- Type
- Mobile app
- PWA: installs on iPhone and Android
- Market since
- 2017
- Established market · AI estimate
ScreensConcept
Mobile app: Sales and technical professionals record and label speakers during calls they take on the go without setup or calendar integration. Sample screens on a phone, in the proposed brand. Their numbers are sample data.
Discovery call
3 speakers
- Dana (you)41% talk time
- Marcus, Acme38% talk time
- Priya, Acme21% talk time
- Duration32 min
- ProcessingOn device
About this appAI proposal
Who Said turns a call into a transcript where each line is tied to the right person. Speaker labelling runs on your device, so audio and text stay with you unless you choose to send them elsewhere. It needs no calendar link and no meeting-bot account.
It is built for technical professionals and salespeople who run discovery calls and care about privacy. When a speaker is split in two or mislabelled, you fix it by tapping a line or saying the correction aloud, and the fix applies to the whole transcript.
An optional paid cloud step adds context-aware speaker merging, cleanup, summaries and action items assigned to the person who took them. It is off by default and chosen per meeting.
- Tagline
- Know who said it. Keep it on your device.
- Voice
- PlainCalmPrecise
- Why this name
- It names the exact question the product answers in two plain words: who said that line.
In the first version
Members say which features they would use. It tells PWAS what people want; it changes no score.
On-device speaker labelling
Voices are separated and labelled on your device. The free tier never uploads audio or text.
No member yet
No calendar required
Press record and go. With no calendar access, past transcripts never clutter your schedule.
No member yet
Merge split speakers
When one person shows up as two speakers, merge them once and every line updates.
No member yet
Voice corrections
Say 'Speaker 2 is Dana' and the transcript is renamed without you stopping to type.
No member yet
Brand identityAI proposal
Check the name against trademark registers before using it.
Palette
#0F4C5CDeep Teal · primary#F6F3ECPaper · surface#1B2A30Ink · text#E0902BAmber · accent#7FA99BSage · data
Typography
Aa
Who Said
ABCDEFGHIJKLMNOPQRSTUVWXYZ 0123456789
Know who said it. Keep it on your device.
PricingAI proposal
The material gives no competitor prices. The free on-device tier and paid cloud tiers follow the opportunity's model; the $15 figure is a starting point to test.
Local
Free
- On-device speaker labelling
- Manual and voice corrections
- Local export and delete
Pro
$15 a month
- Cloud speaker merging and cleanup
- Summaries
- Action items by speaker
- Pick cloud per meeting
Team
Contact us
- Shared glossary of names
- Admin privacy settings
- Volume pricing for sales teams
Would you pay Pro · $15 a month?
Your answer tells PWAS which prices people accept. It changes no score.
No answers yet.
What it could be worthEstimate
What a software business making this much sells for today, from real deals. Pick a monthly revenue.
If it makes
Worth about
$194,000
Sale range $132,000 to $263,000
$120,000 a year in revenue
From 87 software businesses listed and sold on Empire Flippers in the last year, revenue vetted by the broker: they sell for about 19× their monthly revenue, most between 13× and 26×.
Measured on profit, the middle deal is 32× monthly net profit: about 2.7 years of profit.
An estimate, not an offer or financial advice. A real price depends on profit, growth, churn, how much of the work is the owner's, and the buyer.
How it would growAI proposal
How the product would reach new customers through its own use, and how the first ones would be found. A plan to test, not a result.
- Built-in loop
- Shared outputs. Users share transcripts and summaries with speaker names as links or Markdown with clients and colleagues, each carrying a small 'made with Who Said' note.
- Do they become customers?
- Recipients see the product but may not record calls themselves. Test whether colleagues of a sender, such as other salespeople on the same team, start their own accounts from the shared page.
- First customers
- Solo and small-team B2B salespeople and sales engineers who run discovery calls and work with privacy-sensitive buyers.
- Where to reach them
- Hacker News and developer forums: Show a demo of on-device labelling and invite technical users to try it and report accuracy.
- Sales communities: Offer the free tier in sales-ops groups and newsletters, with a discovery-call template.
- Consulting firm outreach: Contact small consultancies directly about private client-meeting notes.
- First milestone
- Ten paying Pro users from one vertical, such as B2B sales, within 60 days of launch, with most meetings labelled correctly.
Why people already pay for this
- VerifiedHacker News post received 32 points and 10 comments
- Public claimThe app achieves 97% accuracy at identifying speakers, presented as a major strength and possibly the first of its kind.
- AI inferenceAccurate speaker identification was a need the builder experienced in previous roles and wanted to solve.
- AI inferenceUsers need to correct speaker attribution and merge incorrectly segmented speakers in transcripts.
- AI inference
Tech stack
Every proposal builds on the same PWA base, installable from the browser on any phone or computer. Dashed boxes are what this app adds.
- AppNext.js and TypeScript
- PWAManifest and service worker
- UITailwind CSS
- Data and authSupabase Postgres
- PaymentsStripe
- EmailResend
- HostingVercel
- This app · On-device speech modelsWebAssembly/WebGPU speaker diarization and transcriptionCore labelling must run in the browser so audio stays on the device.
- This app · Cloud language stepHosted LLM API with no-retention termsPowers optional summaries, merging and action items per meeting.
RisksAI inference
- TechnicalThe 97% accuracy figure is a builder's public claim for another app. Reaching it on-device in a browser, with overlapping speech and poor audio, is unproven.
- PlatformCapturing the other side of a call from a PWA is limited on phones and varies by browser, so some users may only get in-room recording.
- CommercialUsers may accept the free on-device tier and not pay for cloud features, especially privacy-minded ones who distrust cloud steps.
- RegulatoryRecording calls may need consent from other participants depending on location. The product needs clear prompts and cannot give legal advice.
- AICloud summaries can attribute an action item to the wrong person, which hurts trust in a tool sold on correct attribution.
Everything marked AI proposal or AI inference was written by a model from the evidence on the opportunity page. The proposed name must be checked against trademark registers before it is used.