Transcription services cost: build your real budget

Stop shopping transcription on the per-minute rate. The number that decides your budget is total monthly spend for your actual volume, and that flips the ranking of every vendor depending on how much you transcribe. For someone running regular calls, a flat per-seat AI tool (free to about $20 a month) is almost always the cheapest correct option, because per-minute pricing punishes consistency. You only pay real money, roughly $2 per minute, when you genuinely need certified human accuracy on a specific deliverable.

This guide does the math you actually need: worked monthly budgets for the four people who ask this question most (a solo professional, a five-person sales team, an agency billing client meetings, and a one-off legal deposition), plus the surcharges that don't show up on the pricing hero. The goal is not a survey of vendors. It is a number you can put in a spreadsheet today.

The only formula that matters

Every transcription decision comes down to one comparison: your real monthly volume times the per-minute rate, versus a flat per-seat subscription. Write it out before you compare anything.

Monthly transcription budget
----------------------------
A. Hours you transcribe per month       (be honest, count recurring calls)
B. Minutes                = A x 60
C. Per-minute cost        = B x rate     (human ~$1.99, AI credits ~$0.10 to $0.25)
D. Flat per-seat cost     = seats x monthly plan price

Cheaper option = min(C, D)
Break-even hours = (seats x plan price) / (60 x rate)

That last line is the one nobody computes. At a $13-per-seat flat plan and a $0.25-per-minute AI credit rate, the break-even is under one hour a month. Translation: if you transcribe more than about 52 minutes of meetings in a month, the flat plan is already cheaper than buying credits, and it keeps getting cheaper from there. Per-minute math only wins for people who transcribe almost nothing, or who need the human premium on a single file.

$1.99/min

Rev human, 99%+ accuracy

$0.25/min

Rev AI, 96%+ accuracy

$16.99/mo

Otter Pro (1,200 min cap)

$13/seat/mo

Scribbl Pro annual, unlimited

Four worked budgets (find the one that's you)

Here is the actual dollar math for the four situations that drive almost every search for this answer. All rates are verified against vendor pricing pages as of June 2026.

The solo professional: 8 hours of calls a month. A consultant, recruiter, or founder who records their own meetings. At Rev's $0.25-per-minute AI credit rate, 8 hours (480 minutes) is $120 a month. On a flat per-seat AI tool, the same 8 hours costs $13 to $29 a month, and on a free tier built for individuals it can cost $0. The per-minute route here is roughly 6x to 9x more expensive for identical output. Verdict: flat plan, or a free tier if your volume fits.

The five-person sales team: 60 hours a month total. Five reps, a dozen calls each. At $0.25 per minute that is 3,600 minutes, or $900 a month, every month. Five flat seats at $13 to $29 each is $65 to $100 a month, unlimited. The gap is roughly 9x to 14x, and it compounds. This is the single most expensive mistake teams make: buying per-minute credits for recurring sales calls. Verdict: per-seat unlimited, no contest. (If you sell for a living, the sales playbook covers why searchable call transcripts pay for themselves beyond the raw cost.)

The agency billing client meetings: mixed, plus one verbatim deliverable. Say 40 hours of internal and client calls (notes are fine at AI accuracy) plus one 90-minute client interview that has to be a clean published transcript. The 40 hours go on flat per-seat tooling (a handful of seats, well under $100 a month total). The single 90-minute deliverable goes to a human service: 90 minutes times $1.99 is about $179, billed once, ideally passed through to the client as a line item. Mixing the two is the move. Putting all 40 hours through a human service would cost about $4,776 a month. Verdict: flat plan for the bulk, human per-file for the one thing that needs it. The guide for agencies digs into billing this back to clients.

The one-off legal deposition: 2 hours, certified. No recurring need, but the transcript is the deliverable and errors carry consequences. Two hours at Rev's $1.99 per minute is about $239, possibly more with rush turnaround or strict verbatim. Verdict: pay the human rate per minute and do not buy a subscription you will not use again. This is the one scenario where the headline per-minute price is the right thing to shop on.

Per-minute makes sense

• One-off, high-stakes: deposition, published interview
• You need 99%+ human accuracy: legal, medical, broadcast
• Almost no volume: a file or two a year
• Budget as: a per-deliverable line item, often passed to a client

Flat per-seat wins

• Recurring meetings: standups, sales calls, client check-ins
• AI accuracy is fine: internal notes, searchable archive
• Any real volume: break-even is under 1 hour a month
• Budget as: a fixed monthly line you can forecast

The surcharges that wreck a budget

The base rate is the start of the bill, not the end of it. Every one of these is real, and none of them appear on the pricing hero. Price your actual usage pattern, not the sticker.

• Rush turnaround. Standard human pricing assumes you can wait 12 hours or more. Same-day or rush orders cost more, sometimes much more. If the deadline is tight, budget a premium on top of the per-minute rate.
• Strict verbatim. Capturing every "um," false start, and crosstalk takes a transcriber longer than a clean read, and it is priced higher. Decide which you actually need before you order.
• Per-file add-ons. Timestamps, multiple-speaker handling, and burned-in captions are often line items on top of the base rate. GoTranscript, for example, charges extra per file for open captions.
• Minute caps and length limits on AI tools. Otter's free tier caps at 300 minutes a month with a 30-minute-per-conversation limit, and even Otter Pro caps single conversations at 90 minutes. A long workshop can blow past these mid-recording.
• Storage vs transcription minutes. Some tools give you generous transcription but cap how much you can keep. Fireflies' free tier offers unlimited transcription but only 400 minutes of stored audio per team. "Unlimited" sometimes means "unlimited until it expires."
• Your own editing time. Even good AI needs a cleanup pass, and a human transcript needs a read-through before it reaches a client. That is your labor at your hourly rate, and it belongs in the budget.

"Free" native transcription has a price tag too

Before you budget for any tool, check what's bundled in the plans you already pay for. Native transcription in Google Meet, Microsoft Teams, and Zoom can make the marginal cost of transcribing a meeting you attend effectively zero. But "free" here is doing a lot of work, and the catches are real.

So Google Meet transcription is "free" only if your organization already pays for Business Standard or higher, which is a per-seat license cost in its own right. It is also opt-in every single meeting and owned by whoever organized the call, which is a problem if you joined someone else's call or want your own copy. For the step-by-step native path, see getting a Google Meet recording transcript. The admin setup, license seats, manual activation, and organizer-only ownership are all costs that never show up on a "free" label.

What the AI meeting tools actually cost

If native won't cover it, here is what the popular flat-rate tools charge in 2026, with the caps that matter for budgeting.

• Otter.ai. Free with 300 transcription minutes a month and a 30-minute-per-conversation cap. Pro is $16.99 per user a month, or $8.33 billed annually, with 1,200 minutes a month and a 90-minute cap. Business is $30 per user a month, or $19.99 annually, with unlimited minutes and a 4-hour cap.
• Fireflies.ai. Free with unlimited transcription but only 400 minutes of stored audio per team. Pro is $10 per user a month billed annually with 8,000 storage minutes per seat. Business is $19 per user a month annually with unlimited storage.
• Rev. Subscriptions bundle AI minutes per seat (Free gives 45 AI minutes a month; paid tiers add thousands plus a discount on human work), but Rev's real value is the human service at $1.99 per minute when you need certified accuracy.
• Scribbl. Free for up to 10 meetings a month at unlimited length, with AI notes and action items. Pro is $29 per user a month, or $13 billed annually, with unlimited meetings. The desktop app adds Zoom and Microsoft Teams support so you avoid a separate per-platform license.

Otter / Fireflies

• Otter free: 300 min/mo, 30-min-per-call cap
• Otter Pro: $8.33/user/mo annual, 1,200 min
• Fireflies free: unlimited transcription, 400 storage min/team
• Fireflies Pro: $10/user/mo annual, 8,000 storage min

Rev / Scribbl

• Rev human: $1.99/min, 99%+, billed per file
• Rev AI: $0.25/min, 96%+
• Scribbl free: 10 meetings/mo, no length limit
• Scribbl Pro: $13/user/mo annual, unlimited, no bot in the call

How to land on your number

Run these four steps in order. The answer usually falls out by step three, and you will have a defensible line item instead of a guess.

1. 1

   Count your real monthly volume

   Add up recurring calls, not the one busy week. Convert to hours, then minutes. This single number drives everything.

2. 2

   Decide the accuracy stakes

   Internal notes and searchable archive? AI accuracy is fine. A court record, medical chart, or published verbatim quote? Budget the human rate for that file only.

3. 3

   Check what you already own

   On paid Google Workspace, Microsoft 365, or Zoom? Native transcription may already be included. Test it before buying, and weigh the manual-start and organizer-ownership limits.

4. 4

   Run the break-even, then total the TCO

   Compare per-minute volume against a flat seat. Add rush, verbatim, add-ons, storage, and your editing time. The cheapest sticker rarely wins; the right total does.

For most people the answer is a flat per-seat AI tool, because most transcription is routine meeting notes, not certified verbatim. If you live in Google Meet and want capture without a bot announcing itself in the call, Scribbl is free for individuals (10 meetings a month, no length limit) and Pro is a flat $13 to $29 per seat for unlimited meetings, with no per-minute meter. The desktop app adds Zoom and Microsoft Teams, and you can see the tiers on the pricing page. When the deliverable is a deposition or a published interview, skip the tools and pay a human per minute. That is the one place the premium earns its keep. For the no-cost options specifically, the companion piece on the best free transcription software breaks them down by category.

Frequently asked questions

How much do transcription services cost per minute in 2026?

Human transcription runs about $1.99 per audio minute for 99%+ certified accuracy (Rev's rate, delivered in 12 hours or less), so roughly $120 an hour. AI transcription is far cheaper: Rev's AI is $0.25 per minute at 96% accuracy, and flat per-seat AI meeting tools bundle unlimited minutes into $0 to about $30 a month. There is no single rate, because price tracks the accuracy and turnaround you actually need.

Per-minute or flat subscription: which is cheaper for me?

Run the break-even. At a $13-per-seat plan and a $0.25-per-minute credit rate, you pass break-even at under an hour of transcription a month. Above that, flat unlimited wins, and the gap widens fast. A five-person team running 60 hours a month pays about $900 on per-minute credits versus $65 to $100 on flat seats. Per-minute only wins for near-zero volume or a single high-stakes file.

Is AI transcription accurate enough to skip human transcribers?

For internal use, yes. AI typically lands around 90 to 96% accuracy, which is plenty for meeting notes, sales calls, action items, and a searchable archive. It is not enough when the transcript is the deliverable and errors carry consequences: legal records, medical charts, regulatory filings, or published verbatim quotes. For those, the human premium (8x to 10x the AI price) is insurance, not waste. Match the accuracy to the stakes.

Do I have to pay extra to transcribe meetings I already host?

Often no. Paid Google Workspace, Microsoft 365, and Zoom plans include native transcription, so the marginal cost of a meeting you attend can be effectively zero. But native features are gated to the host's license, must be started manually, save to the organizer's Drive or OneDrive, and offer no unified archive across platforms. Those limits are exactly why dedicated AI tools exist. Test native first, then decide.

How should an agency budget transcription for client work?

Split it. Put the bulk of internal and client meeting notes on flat per-seat AI tooling (a few seats, well under $100 a month), and reserve the human per-minute rate for the specific deliverables that need certified accuracy, billing those back to the client as a line item. Putting every hour through a human service is how agencies turn a $100 budget into a $5,000 one. See the guide for agencies for the full breakdown.