Maya K.
Senior Full-Stack Developer
Human transcription costs $1.00 to $3.00 per audio minute in 2026, which is $60 to $180 per audio hour. Automated transcription runs $0.05 to $0.25 a minute. Billed by working time instead, transcriptionists charge $20 to $50 an hour.
Transcription is quoted in six different units, which is why two quotes for the same file can look nothing alike. Three fixed conversion factors reconcile all of them. People speak at about 150 words a minute. A standard page is 250 words and a medical line is 65 characters. And one audio hour takes a trained transcriptionist 4 to 6 working hours.
Convert everything and four independently compiled price lists land on the same number. General per-minute rates give $60 to $180 an audio hour. Medical per-line rates of $0.09 to $0.18, across the 831 lines in an audio hour, give $74.77 to $149.54. Federal courts cap an ordinary transcript at $4.40 a page, which at 25 to 40 pages an hour of proceedings is $110 to $176. And Rev, one of the largest providers, publishes $1.99 a minute, or $119.40 an audio hour. Nobody coordinated those four lists.
The same arithmetic run backwards shows where the floor is. Divide an audio-hour price by the 4 to 6 hours the work takes and you get the wage inside it. At $0.72 a minute, the bottom of the published human band, a 6:1 file pays $7.20 an hour, below the $7.25 federal minimum wage.
Find your freelancer
PreviewYour top matches
2 of 3 can start this week All 3 available this month Flexible start dates
Senior Full-Stack Developer
+ 2 more on a full shortlist
Product Designer
+ 2 more on a full shortlist
Conversion Copywriter
+ 2 more on a full shortlist
Growth Marketer
+ 2 more on a full shortlist
Video Editor
+ 2 more on a full shortlist
Small-Business Bookkeeper
+ 2 more on a full shortlist
Executive Virtual Assistant
+ 2 more on a full shortlist
Technical SEO Consultant
+ 2 more on a full shortlist
Google Ads Consultant
+ 2 more on a full shortlist
Meta Ads Consultant
+ 2 more on a full shortlist
Senior Data Scientist
+ 2 more on a full shortlist
Senior DevOps Engineer
+ 2 more on a full shortlist
Social Media Manager
+ 2 more on a full shortlist
Management Consultant
+ 2 more on a full shortlist
Freelance CPA
+ 2 more on a full shortlist
Direct-Response Copywriter
+ 2 more on a full shortlist
Senior QA Automation Engineer
+ 2 more on a full shortlist
Certified Salesforce Developer
+ 2 more on a full shortlist
Senior UX Designer
+ 2 more on a full shortlist
Senior Webflow Developer
+ 2 more on a full shortlist
Certified Legal Translator
+ 2 more on a full shortlist
Senior Commercial Illustrator
+ 2 more on a full shortlist
Fund milestone 1 · $1,800 in escrow · $4,800 in escrow · $12,500 in escrow · released on your approval
Price the work, not the pitch
Every specialist we match shows a rate on the card before you talk, so you can price a file against the tables below instead of waiting on a quote. Milestone escrow holds the money until the transcript lands and you have read it.
Last updated August 2026
By the audio minute
Per audio minute is the standard unit because the recording length is the one quantity both sides can check before anyone starts. Two things move the number, and neither is the subject matter on its own: how many people are talking, and how good the recording is. A single speaker into a lapel mic and a four-person round table in a glass conference room are different jobs even when the words are identical.
| Recording type | Per audio minute | Per audio hour | Why it sits there |
|---|---|---|---|
| Automated, no human pass | $0.05 to $0.25 | $3 to $15 | Machine output, roughly 94% to 96% on clean audio |
| Single speaker, clear audio, standard turnaround | $1.00 to $1.75 | $60 to $105 | The easiest human job: one voice, no crosstalk |
| Two-person interview | $1.50 to $2.25 | $90 to $135 | Speaker labeling plus some overlap to untangle |
| Focus group or panel, three or more voices | $2.50 to $3.00 | $150 to $180 | Crosstalk forces repeated replays of the same seconds |
| Poor audio, heavy accents, background noise | $2.50 to $5.00 | $150 to $300 | Ratio climbs past 8:1, so the labor doubles |
| Legal: hearings, depositions, recorded interviews | $1.50 to $3.00 | $90 to $180 | Formatting standards and certification requirements |
| Medical dictation and clinical notes | $1.25 to $2.50 | $75 to $150 | Usually billed per 65-character line rather than per minute |
| Certified court reporter, live capture | $3.00 to $5.00+ | $180 to $300+ | A licensed officer attending in person, plus appearance fees |
Use $1.75 a minute as your sanity check for ordinary business audio. Multiply it by your recording length before you request a single quote and you will know immediately whether a bid is a discount or a different product. Anything under about $0.72 a minute is either machine output or labor priced outside the US market, and the arithmetic further down explains why there is no third option.
The conversion
This single ratio explains almost every argument about transcription pricing. A trained transcriptionist needs 4 to 6 hours to finish one hour of audio, and someone without training needs 5 to 8 or more. The cause is mechanical rather than motivational. People speak at 125 to 175 words a minute. Most typists sustain 60 to 90. The audio therefore has to be stopped, rewound and replayed continuously, and then checked again for names, figures and anything the ear guessed at.
Once you accept the ratio, you can price any quote in the only terms that matter: what it pays the person doing the work. Take the per-audio-hour figure and divide it by 4, 5 or 6. That is the hourly wage buried inside the rate, and it is the number that tells you whether the transcript will be done by a professional, by a beginner, or by a machine with a human name on the invoice.
| Rate | Per audio hour | At 4:1 | At 5:1 | At 6:1 |
|---|---|---|---|---|
| $0.72 / min | $43.20 | $10.80 | $8.64 | $7.20 |
| $1.00 / min | $60.00 | $15.00 | $12.00 | $10.00 |
| $1.50 / min | $90.00 | $22.50 | $18.00 | $15.00 |
| $1.75 / min | $105.00 | $26.25 | $21.00 | $17.50 |
| $1.99 / min (Rev) | $119.40 | $29.85 | $23.88 | $19.90 |
| $2.25 / min | $135.00 | $33.75 | $27.00 | $22.50 |
| $3.00 / min | $180.00 | $45.00 | $36.00 | $30.00 |
The floor
The bottom row of the published human band, $0.72 an audio minute, works out to $43.20 an audio hour. On a difficult file at a 6:1 ratio that is $7.20 for an hour of concentrated work, which is less than the $7.25 federal minimum wage, unchanged since 24 July 2009. Nobody in the United States does careful transcription at that price for long. When you see it quoted, the work is being done somewhere else, by a machine, or by someone who has not yet worked out what the job takes.
The cross-check
General transcription, medical transcription, federal court reporting and one large commercial vendor all publish prices in different units, for different buyers, with no reason to match each other. Convert them into the same unit and they land on top of each other. That agreement is the strongest evidence available that the market rate for an hour of human transcription is real and not a negotiating position.
| Price list | As published | Conversion used | Per audio hour |
|---|---|---|---|
| General commercial transcription | $1.00 to $3.00 per audio minute | 60 minutes in an hour | $60.00 to $180.00 |
| Medical transcription, per line | $0.09 to $0.18 per 65-character line | 831 lines per audio hour | $74.77 to $149.54 |
| Federal courts, ordinary transcript | $4.40 maximum per page | 25 to 40 pages per hour of proceedings | $110.00 to $176.00 |
| Rev, published human rate | $1.99 per audio minute | 60 minutes in an hour | $119.40 |
| Hourly-billed transcriptionist time | $20 to $30 per working hour | 4 to 6 working hours per audio hour | $80.00 to $180.00 |
The clincher is the middle of the range rather than the edges. Rev publishes $119.40 an audio hour. The midpoint of the general band is $120.00. The two differ by sixty cents, and they were compiled by people with no reason to agree. If a quote for ordinary business audio sits far outside $90 to $120 an audio hour, the difference is in the product, not in the pricing.
The other units
Three industries bill transcription by output rather than by recording length, and each defines its unit differently. A page in a legal transcript is not a page in a general document, and a line in medical transcription is not a line of text on your screen. Getting these definitions wrong is the most common way a transcription invoice comes in higher than the quote implied.
| Unit | How it is defined | Typical rate | Per audio hour |
|---|---|---|---|
| Audio minute | 60 seconds of recording, not of work | $1.00 to $3.00 | $60 to $180 |
| Working hour | An hour of the transcriptionist's time | $20 to $50 | $80 to $300 |
| Standard page | 250 words, about 100 seconds of speech | $1.65 to $5.00 | $60 to $180 |
| Legal transcript page | 25 double-spaced numbered lines | $3.00 to $6.00 | $75 to $240 |
| Medical line | 65 characters including spaces (AHDI standard) | $0.09 to $0.18 | $75 to $150 |
| Per word | Rare in transcription, common in translation | $0.007 to $0.02 | $60 to $180 |
The 65-character line
Medical transcription bills per line, and a line is 65 characters including spaces and punctuation, not a line as it appears on a page. One audio hour of dictation runs to roughly 831 lines. Ask any medical vendor to confirm the character count and whether headers, blank lines and signature blocks are counted, because those choices move a monthly invoice by double digits. Our medical transcription services pricing guide converts that line rate into a per-minute one and shows where the two units stop agreeing.
The 25-line page
A legal transcript page is 25 numbered, double-spaced lines with wide margins, which is why an hour of testimony fills 25 to 40 of them. That format is a standard, not padding: it exists so that counsel can cite a page and line number and everyone finds the same words. It also means legal per-page pricing is not comparable to general per-page pricing.
Transcription against translation
At $1.50 a minute and 150 spoken words a minute, transcription costs about $0.01 a word. The going rate for translation is about $0.16. Transcribing a recording costs roughly a sixteenth of what translating the same words costs, which is why any subtitling quote is dominated by the translation line, not the transcript.
Machine against human
Rev sells both products and prices them at $0.25 and $1.99 an audio minute, a spread of 7.96 times for the same recording. That gap is the most important number in transcription buying right now, and it is not as generous as it looks, because the cheaper product arrives unfinished. The question is never whether the machine is accurate. It is who fixes the part it got wrong, and what their hour costs.
| Source | Typical accuracy | Wrong words per audio hour | What that looks like in practice |
|---|---|---|---|
| Professional human, clear audio | 99%+ | 90 or fewer | Occasional proper nouns, usually flagged as inaudible |
| Best automated engines, clear audio | about 96% | 360 | One word in 25, silently wrong rather than marked |
| Mainstream automated tools, clear audio | about 94% | 540 | Names, figures and technical terms are the first casualties |
| Automated, noisy or multi-speaker audio | about 86% | 1,260 | Speaker attribution collapses, whole exchanges merge |
The counts come from a simple conversion. Speech at 150 words a minute puts about 9,000 words in an audio hour, so a 4% error rate is 360 wrong words to find. The finding matters more than the percentage, because a machine error is not marked. A human transcriptionist writes "[inaudible 14:22]" and you know where to look. An engine produces a confident, grammatical, wrong sentence, and the only way to find it is to listen to the whole recording against the transcript.
| Route | Direct cost | Internal review | True cost per audio hour |
|---|---|---|---|
| Automated only, published as final | $15.00 | none | $15.00 |
| Automated plus a 1 hour skim | $15.00 | $27.78 | $42.78 |
| Automated plus a 1.5 hour check against audio | $15.00 | $41.67 | $56.67 |
| Automated plus a full 2 hour verification | $15.00 | $55.56 | $70.56 |
| Finished human transcript | $90 to $120 | spot check only | $90 to $120 |
Review time is costed at $27.78 an hour, the loaded cost of a US employee earning the median transcriptionist wage. The honest conclusion is that automation does save real money, but the saving is roughly two to one rather than eight to one once a person checks the output. The eight to one figure only holds when nobody checks, which is a defensible choice for internal search and a poor one for anything a client, a regulator or a court will read.
Line items
Headline transcription rates are quoted for the easiest version of the job. Every add-on below is standard, published and legitimate, but they stack, and four of them together can double the advertised rate. Ask which are included before you compare two quotes, because vendors differ in what they bundle rather than in what they charge.
Extra speakers
About $0.25 a minute for 3 or more
Labeling who said what, and untangling people talking over each other, is the single biggest driver of transcription labor. Some vendors reprice the whole file into a higher tier rather than adding a surcharge.
Full verbatim
Plus $0.25 to $0.50 a minute
Every false start, repetition and filler word, captured exactly. Research, litigation and linguistic work need it. Marketing and internal notes almost never do, and paying for it by accident is a common overspend.
Timestamps
Plus $0.25 to $0.50 a minute
Frequent timestamps cost more than sparse ones, and per-speaker-change stamps cost more than fixed intervals. Decide the interval up front; adding stamps afterwards means reworking the file against the audio.
Rush turnaround
Plus 50% to 100%
A one to two day turnaround typically costs half again as much as a three to five day standard, and same-day can double it. One published tier runs $1.50 extended, $1.75 standard and $2.25 rush for identical work.
Audio quality
Often a whole tier, not a surcharge
Phone recordings, echoing rooms and distant microphones push the ratio past 8:1. A $30 lapel mic on the main speaker will save more on a year of transcription than any amount of negotiating over the rate.
Certification and notarization
Free to $150 flat
Some legal providers include a certificate of accuracy at no charge. Notarization is separate and commonly $150 as a flat fee. Confirm which your filing actually requires before paying for both.
The cliff
Most professional transcription providers apply a minimum project fee, commonly around $100. It exists for a defensible reason: intake, file handling, quality review and invoicing cost the same whether the recording is four minutes or forty. But it means the advertised per-minute rate is fiction on anything short, and buyers who send files one at a time pay it repeatedly.
At a $1.75 list rate, a $100 minimum stops binding at 57.1 audio minutes. Below that you are paying more than list, and the shorter the file the worse it gets. The fix costs nothing: batch short recordings into a single order, and ask whether the minimum applies per file or per submission before you send anything.
| File length | Charged | Effective per minute | Versus $1.75 list |
|---|---|---|---|
| 3 minutes | $100.00 | $33.33 | 19.0x |
| 5 minutes | $100.00 | $20.00 | 11.4x |
| 10 minutes | $100.00 | $10.00 | 5.7x |
| 20 minutes | $100.00 | $5.00 | 2.9x |
| 30 minutes | $100.00 | $3.33 | 1.9x |
| 57 minutes | $100.00 | $1.75 | 1.0x |
| 90 minutes | $157.50 | $1.75 | 1.0x |
The same minimum at a $1.50 rate binds all the way to 66.7 minutes, and at a $2.25 rush rate only to 44.4. A lower advertised rate therefore extends the zone where the advertised rate does not apply, which is worth knowing before you pick a vendor on headline price alone.
In-house
The Bureau of Labor Statistics puts the median wage for medical transcriptionists at $40,410 a year, or $19.43 an hour, across 43,900 people, with employment projected to decline through 2034 and about 7,400 openings a year. Loaded with benefits at the BLS employer-cost ratio, where wages are 69.9% of total compensation, that salary costs $57,786.30, or $27.78 an hour across 2,080 paid hours.
Here the 4:1 ratio does something it does not do in any other role we have priced. It caps how much audio one person can physically process in a year, and that cap sits below the point where in-house becomes cheaper. Even if the hire spends every one of their 2,080 paid hours transcribing and nothing else, they clear only 520 audio hours a year.
| Scenario | Hours on transcription | Audio hours cleared | Cost per audio hour |
|---|---|---|---|
| Full time, easy audio (4:1) | 2,080 | 520.0 | $111.13 |
| Full time, mixed audio (5:1) | 2,080 | 416.0 | $138.91 |
| Full time, hard audio (6:1) | 2,080 | 346.7 | $166.69 |
| Half the week on transcription (4:1) | 1,040 | 260.0 | $222.25 |
| A day a week on transcription (4:1) | 416 | 104.0 | $555.64 |
| Outside service, standard business audio | none of yours | unlimited | $90 to $120 |
Read the top row again, because it settles the question. A dedicated full-time transcriptionist working on nothing else, on the easiest possible audio, costs $111.13 an audio hour. An outside service charges $90 to $120. The in-house option does not lose narrowly at low volume and win at high volume the way most build-versus-buy comparisons do. It never wins, because the labor ratio is the same for your employee as for the vendor, and the vendor has no idle time to carry.
There are still three good reasons to hire: confidentiality that no vendor agreement satisfies, institutional knowledge of names and terminology that pushes accuracy above what any outsider reaches, and transcription being one part of a broader role rather than the whole job. Cost is not on that list. If the case for the hire rests on saving money, the arithmetic above is the answer.
By budget
$3 to $15
Automated only
A searchable draft with 360 to 540 silent errors an hour. Genuinely useful for finding where in a recording something was said, and for internal notes nobody will quote. Unsuitable for anything published, filed or shown to a client.
$40 to $70
Automated plus your own review
The real cost of the cheap route once someone on your payroll checks it against the audio. Works when the reviewer already knows the material, and stops working the moment the file is long, noisy or unfamiliar.
$60 to $105
Human, single speaker, standard turnaround
Clean audio, one voice, three to five business days. A finished transcript with inaudible sections marked rather than guessed. This is the right default for podcasts, webinars, dictation and one-to-one interviews.
$90 to $135
Human, two speakers, labeled
Interviews and user research, with speaker attribution you can rely on when quoting. Add timestamps here rather than later, since retro-fitting them means a second pass against the audio. Transcribing interviews before handing them to a writer is also the cheapest way to cut a book budget, since ghostwriting costs roughly forty times as much per word.
$150 to $180
Human, focus group or difficult audio
Three or more voices, crosstalk, and a room that was never set up for recording. The premium is labor, not margin: these files genuinely take eight or more working hours per audio hour.
$180 to $300+
Certified legal capture
A licensed court reporter attending in person, plus appearance fees and per-page transcript charges. Required for the official record in a deposition, and overkill for transcribing a recording you already hold. Where a witness or patient does not speak English, that is a separate live cost again, priced per minute or against an hourly minimum in our breakdown of video remote interpreting and on-site interpreter rates.
One decision governs most of this: whether anyone outside your team will read the words. Internal search, personal notes and rough recall are well served by the automated tier. Anything quoted in a report, filed with a court, published as a caption or shown to a customer needs a human pass, and the difference between those two worlds is about $75 an audio hour.
Questions people actually ask
Professional human transcription costs $1.00 to $3.00 per audio minute in 2026, which is $60 to $180 per audio hour. Clean single-speaker audio on a standard three to five day turnaround sits at the bottom of that band. Multi-speaker recordings, poor audio and rush delivery push a file toward the top of it.
Expect $1.00 to $3.00 per audio minute for human transcription and $0.05 to $0.25 for automated. Rev prices the two side by side at $1.99 and $0.25, a spread of almost eight times. Below about $0.72 a minute you are no longer buying US human labor, whatever the listing says.
One audio hour costs $60 to $180 with a human transcriptionist and $3 to $15 with automated software. The wide human band reflects difficulty, not vendor greed: an audio hour is 4 to 6 working hours of labor, so a $90 transcript pays the person doing it $15 to $22.50 an hour.
An experienced transcriptionist needs 4 to 6 hours to transcribe one hour of audio. Without training it takes 5 to 8 hours or longer. The gap exists because people speak at 125 to 175 words a minute while most typists manage 60 to 90, so the audio has to be paused and replayed constantly.
Roughly 2 to 3 working hours for a trained transcriptionist on clear audio. Add an hour or more if three or more people are speaking over each other, the recording was made in a room with hard surfaces, or the content is dense with names, drug names, part numbers or case citations.
Freelance transcriptionists charge $1.00 to $3.00 per audio minute, or $20 to $50 per hour of their own working time when they bill hourly. The two units are closer than they look. At a 5:1 ratio, $20 an hour of working time equals $1.67 per audio minute.
It depends which page is meant. A standard 250-word page holds about 100 seconds of speech, so at $1.50 an audio minute it costs about $2.50. A legal transcript page is 25 double-spaced lines and is billed at $3 to $6 by court reporters, with federal courts capping an ordinary transcript at $4.40 a page.
Medical transcription is billed at $0.09 to $0.18 per 65-character line, most commonly $0.10 to $0.16. The 65-character line including spaces and punctuation is the AHDI standard. One audio hour of dictation runs to roughly 830 such lines, so the per-line price implies $75 to $150 an audio hour.
An audio hour is 60 minutes of recording length, not 60 minutes of anyone working. It is the billing unit because it is the only quantity both sides can verify before the job starts. The labor inside one audio hour is 4 to 6 working hours, which is what the price actually pays for.
The best automated engines reach about 96% on clear audio against 99% or better for a human. On 9,000 spoken words in an audio hour that is 360 wrong words versus 90. Accuracy falls another 8 to 12 points on noisy or multi-speaker recordings, where a machine transcript can carry over 1,200 errors an hour.
Only if the vendor signs a business associate agreement and you use the tier the BAA covers. Many consumer transcription apps are not covered on their free or entry plans even when the paid enterprise tier is. Ask for the signed BAA and the plan it applies to before any protected health information is uploaded.
For general US business content on a standard turnaround, $1.50 to $2.00 per audio minute is the going rate in 2026, or $90 to $120 an audio hour. Rev, one of the largest providers, publishes $1.99. Treat $1.75 as your sanity check and price your files against it before requesting quotes.
Tell us the recording length, how many people are speaking and the turnaround you need, and we hand-pick a shortlist of vetted specialists within 48 hours, each with their rate on the card. Milestone escrow holds the money until the transcript lands.