Six tools that turn a client meeting into text. They are not interchangeable: two keep the audio on your own laptop, three send it to a company's servers, and one is built by the government for its own officers. Sort by that column first, price second.
Transcribe
GovTech Singapore · transcribe.gov.sg
Government-built speech to text, for public officers.
What it isA web app and an API that turns live meetings or uploaded audio and video into text, built by GovTech and hosted on the Government on Commercial Cloud.
Security postureCleared to host data classified up to RESTRICTED and SENSITIVE (NORMAL), which no commercial tool on this list claims.
LanguagesFour national languages: English, Chinese, Malay, Tamil.
FeaturesLive and batch transcription, concurrent viewing during a live session, audio-text sync word by word, AI summaries and meeting minutes, shared workspaces with access rights, API access through PlatformAI.
PriceFree to public officers. Login is via the Whole-of-Government Azure Active Directory, or OTP for government domains outside AAD.
EligibilityDesigned for public officers, not private-sector staff. Check with the product team before assuming SME Centre advisors can use it.
Honest note: this is the strongest privacy option on paper and the weakest on eligibility. SME Centre advisors are not public officers, so treat it as "ask first", not "use it on Monday".
Meetily
Zackriya Solutions · MIT licence · macOS, Windows, Linux
Records and transcribes entirely on your own machine. No bot joins the call.
PrivacyTranscription runs locally with Whisper or Parakeet models. Recordings and audio stay on the device. Summaries run either on a local model or through an API key you supply.
Works withZoom, Microsoft Teams, Google Meet, Webex and anything else, because it captures system audio rather than joining as a participant.
Free tierCommunity Edition, open source: real-time transcription, audio and video import in 10 formats, retranscription, AI summaries, 6 built-in summary templates, search and replace, copy and Markdown export.
Pro, US$10/user/monthBilled annually at US$120. Adds speaker identification, custom AI model connector, custom summary templates, auto-detect meetings, PDF and DOCX export, and the MCP server, webhooks and CLI that let another assistant read your meetings.
Try before buying14-day Pro trial, no card. 30-day money-back guarantee. One licence per device.
Reality check31,300 GitHub stars and about 599,000 downloads, actively developed (v0.4.1 in September 2026). Chat with your meetings and calendar integration are still on the roadmap. Linux is a Docker or source build, not a native app.
OpenWhispr
OpenWhispr · open source · macOS, Windows, Linux, iOS
The same local-first promise, with more of the features already shipped.
PrivacyLocal-first. Whisper or Parakeet models run on your device, and speaker diarization is local too.
What it adds over MeetilyGoogle Calendar sync, chat across all your notes and transcripts, agent mode for drafting follow-ups, and push-to-talk dictation into any application.
PriceFree forever: unlimited local transcription, local diarization, calendar sync, 5 hours of meeting recordings a month, dictation, 100+ languages. Pro US$8/month for 20 hours and cloud transcription. Business US$16/month for unlimited recordings and chat over your data.
Where it is thinnerSmaller project than Meetily (about 8,900 GitHub stars), and it is a meeting recorder plus dictation, not a meeting platform integration story.
Conflict you should know about: the detailed OpenWhispr comparison page is written by OpenWhispr, and it still describes Meetily's diarization and chat as roadmap items. Meetily has since shipped diarization in Pro. Read vendor comparisons with the vendor's own interest in mind, including this one.
Otter.ai
Otter · cloud SaaS
The best-known meeting bot. A participant called "Otter" joins your call.
Where the audio goesTo Otter's cloud. Your client's voice is processed on their servers.
Free tier300 transcription minutes a month, 30 minutes per conversation, 3 lifetime audio or video imports.
Pro, US$8.33/month annual(US$16.99 monthly) 1,200 recording minutes a month, 90 minutes per meeting, 10 imports, advanced AI workflows.
Business, US$19.99/month annual(US$30 monthly) Unlimited meetings, up to 4 hours each, unlimited imports, custom workflows, usage analytics.
Best forTeams that want transcripts, action items and a searchable archive without thinking about it. Not for client material you would not email to a vendor.
Fireflies.ai
Fireflies · cloud SaaS
Meeting capture plus analytics, built for teams and sales floors.
Where the audio goesFireflies' cloud. It joins your calls as a participant.
FreeUnlimited transcription and summaries, with 400 minutes of storage for the whole team.
PaidPro US$10/seat/month billed annually (US$18 monthly); Business US$19 (US$29 monthly); Enterprise US$39 annual only. AI features run on credits, priced per query.
StrengthsConversation intelligence, team analytics, broad integrations, multi-language mode on Business.
The catchPer seat, cloud, and the AI credits make the real bill harder to predict than the headline seat price.
HappyScribe
HappyScribe · cloud SaaS
Transcription and subtitling with a human option when it has to be perfect.
Where the audio goesTheir cloud, or a human transcriber for the professional service.
AI pricingFree trial of 10 minutes, then Pro US$29/month for 600 minutes or Business US$89/month for 6,000 minutes, with extra minutes at US$0.20. Pay-as-you-go works out around US$12 per 60 minutes.
Human pricingFrom US$2.00 per minute, 99% accuracy, delivery in 4 hours or more, priced by language and urgency.
Languages150+ for AI, with translation and subtitles.
Best forA recording you already have that must be right: a workshop, a panel, a session you are publishing. Not a live meeting assistant you forget about.
Why this page exists in a talk about assistants
An assistant is only as useful as what it can read. The meeting is where the client's actual pain comes out, and it is the hardest material to capture safely.
Cloud transcribers are convenient and they are the easiest way to move client personal data off the machine without thinking about it. Local transcribers are slightly less polished and they remove that entire question.
For a client-facing advisor in Singapore, that is not a philosophical preference. It is the difference between a tool you can defend in a PDPA conversation and one you cannot.