# Rev.com Pay: What 'Up to $1/Minute' Actually Nets Per Hour

Piper Bowen · August 23, 2026

> Rev.com Pay: What 'Up to $1/Minute' Actually Nets Per Hour. The conversion Rev never prints sits at the midpoint of its own banner: $...

| Takeaway | Detail |
| --- | --- |
| Rev's banner rate is priced per audio minute, not per worked hour | The advertised $0.30–$1.10 range equals $18.00–$66 per audio hour only at a 1:1 completion ratio; at the realistic 3:1–6:1 pace, realized hourly pay falls by the same factor. |
| Even the ceiling misses high-wage state minimums | The $1.10 per-minute top implies $66.00 per audio hour, yet that ceiling cannot reach California's 2026 minimum-wage floor on anything harder than clean audio. |
| Fixed per-minute rates are losing purchasing power | US inflation stood at 3.4% by BLS CPI (July 2026 reading) and 3.7% by BEA PCE (June 2026 reading) — the annual erosion applied to any unchanged per-minute rate through 2026. |
| AI token prices are not transcription rates | Listings spanning $0.96–$3 per 1M tokens describe MiniMax M3 LLM API output pricing on OpenRouter — an off-thesis compute benchmark that should never be attributed to Rev.com. |

The conversion Rev never prints sits at the midpoint of its own banner: $0.60 per audio minute — the center of the advertised $0.30–$1.10 range — is $36.00 per audio hour of source media. Transcriptionists are paid for the audio, not the clock, and clearing one hour of tape takes roughly four hours of work at the industry-standard 4:1 completion ratio. That lands the midpoint at $9.00 per worked hour.

The tax bill narrows the margin further. At the 15.3% self-employment rate, $9.00 becomes $7.62 — 37 cents above the federal minimum wage and $9.28 below California's 2026 floor. The banner's edges offer little shelter: the $0.30 floor implies $18.00 per audio hour and the $1.10 ceiling implies $66.00, but both hold only at a 1:1 ratio and deflate across the realistic 3:1–6:1 spread transcribers actually face.

Context sharpens the warning. US inflation was running at 3.4% by BLS CPI (July 2026 reading) and 3.7% by BEA PCE (June 2026 reading), steadily eroding any per-minute rate left unchanged. And token-price tables quoting $0.96–$3 per 1M tokens are MiniMax M3 LLM API costs from OpenRouter — AI-compute benchmarks, not Rev transcription rates. What 'up to $1/minute' actually nets, in the end, hinges on the ratio Rev never prints.

![Rev.com Pay](https://static.mm-ais.com/article-images-ai/rev-com-pay-what-up-to-1-minute-actually-ai-0909401a.jpg)

## Inside Rev's Meter

The pitch that circulates through every gig-work forum — "Rev pays up to $1.10 per minute, so a fast typist earns $66/hour" — fails on units before it fails on arithmetic. Rev's published 2026 rate card prices source media, not keystrokes: $1.10 buys one minute of audio, not one minute of labor. Even a flawless 2:1 completion on that top-band file yields $33.00 gross per worked hour — about $28 net once the self-employment tax is stripped — and audio pristine enough to sustain 2:1 rarely reaches the freelance board at all.

The mechanism is the meter itself. Rev bills clients per media minute and pays freelancers a fixed per-minute slice, so gross pay locks at the moment of claim: a 60-minute file offered at $0.60/minute pays $36.00 whether production takes 3 hours or 7. The client's invoice never moves; every extra minute of slow audio comes out of the worker's side alone. It is the structural inverse of a W-2 job — revenue fixed, productivity risk delegated entirely to the contractor.

The meter has two dials. The transcription track and the captioning track carry separate pay schedules and separate quality rubrics — captioning layers synchronization and line-break constraints on top of verbatim accuracy. Each returned file is graded, and scores below Rev's threshold lock a freelancer out of higher-paying postings. The top of the $0.30–$1.10 band is therefore not an offer but a conditional, reserved for freelancers whose sustained grades keep the gate open.

Supply is un-metered too. Files surface on a first-come job board with no guarantee of future work, and completed jobs pay weekly via PayPal with no minimum balance. There is no hourly retainer, no shift, no guaranteed volume anywhere in the system — the meter runs only when you claim.

Then the tax wedge. Rev freelancers are 1099 independent contractors: nothing is withheld, and the worker later owes the 15.3% self-employment tax — both the employee and employer halves of Social Security and Medicare, per the IRS's SE tax schedule — on net earnings. Every honest wage comparison therefore runs gross per-minute figures through the 0.847 factor before holding them against any state's minimum wage.

The evidence behind this guide's conclusion comes from three kinds of sources with no reason to agree: the company publishing the rates, the governments setting the floors, and the workers keeping their own tallies. They converge anyway — and agreement between interested and disinterested parties is the strongest signal a wage analysis can produce.

| Completion ratio | Worked hours per 60 audio min | Gross per worked hour | Net after ×0.847 | Verdict |
| --- | --- | --- | --- | --- |
| 3:1 (clean audio) | 3.0 | $12.00 | $10.16 | Above federal floor; below high-wage states |
| 4:1 (mixed audio) | 4.0 | $9.00 | $7.62 | Clears the $7.25 federal floor only |
| 6:1 (crosstalk-heavy) | 6.0 | $6.00 | $5.08 | Below the $7.25 federal floor |

The primary document is Rev's own careers page: it advertises the $0.30–$1.10 per-audio-minute band for transcriptionists quoted throughout this guide, with captioners on a separate tier at $0.54–$1.10 per video minute. Every figure on that page is priced per media minute — sixty minutes of source audio or video — never per clock-hour of labor, and the entire analysis hinges on that unit distinction. The captioner floor sitting far above the transcriptionist floor is the tell: syncing text to a moving video track costs more effort per media minute than transcribing it, and Rev prices that difficulty in. Nowhere does the page quote an hourly figure.

![Pale dawn light filtering through sheer curtains into](https://static.mm-ais.com/article-images-ai/rev-com-pay-what-up-to-1-minute-actually-ai-e2f342f6.jpg)
Pale dawn light filtering through sheer curtains into

## The Evidence

The conversion factor that turns media minutes into worked hours comes from the industry's own training bodies: the Association for Healthcare Documentation Integrity (AHDI) teaches that one audio hour takes roughly four worked hours for a competent transcriptionist — a 4:1 ratio — stretching to 6:1 or worse on poor audio. No accredited curriculum trains anyone to transcribe at playback speed — the reason the fast-typist fantasy collapses on contact with a media-minute rate card.

Finally, the technology ceiling explains why the hardest files stay human. According to OpenAI's Whisper paper (Radford et al., 2022), the system achieves roughly 2–5% word error rate on clean read speech but degrades past 10–20% WER on noisy, multi-speaker recordings. In acoustic terms the failure is categorical: past roughly 10% WER, an automatic transcript stops being a draft to polish and becomes a reconstruction to verify against the audio — often slower than typing from scratch. Automation absorbs the clean, fast files and leaves the board weighted toward crosstalk-heavy jobs where humans operate at 6:1. The ceiling does not threaten Rev's low effective wages; it structurally protects them.

The assembled ledger:

Before claiming anything, find your state's row first: the comparator, not the card, decides whether a file clears your floor.

One number decides every row in this comparison: net effective hourly — (media minutes × offered per-minute rate ÷ measured completion hours) × 0.847. The 0.847 strips the 15.3% self-employment tax, which a contractor owes on both the employer and employee halves of payroll tax; everything else on a rate card is marketing until it passes through this formula. Per-minute rates and per-audio-hour grosses are not comparable across options, because each omits the completion ratio — the multiplier between media minutes and wall-clock hours — which spans 1.5:1 to 7:1 across the tracks below. Two offers at the same per-minute rate can differ by a factor of two in actual pay per hour worked. Any reading of these cards that assumes 1:1 — that a per-media-minute rate is a per-worked-minute wage — dies right here: no track in the table runs at 1:1, and even the fastest, post-editing, runs 1.5:1.

Row 1, Rev's transcription track: at the observed midpoint of about $0.60 per audio minute and a 4:1–6:1 completion ratio, the formula returns $6.00–$9.00 pre-tax, $5.08–$7.62 net. Its advantage is zero-barrier entry — headphones and a laptop — and its disadvantage is a hard rate ceiling. Row 2, captioning: $0.54–$1.10 per video minute looks strictly better, but the deliverables — frame-accurate timing, characters-per-line limits — push ratios toward 5:1–7:1, netting roughly $7–$11 per hour. Higher gross per media minute, similar effective hourly: the stricter format taxes back what the higher rate grants.

Row 3 is the only structural escape: post-editing an ASR draft instead of transcribing from a blank page. The machine absorbs first-pass decoding, cutting the ratio to about 1.5:1–2:1 and lifting effective hourly to $12–$20 wherever such volume exists — unsurprising from the speech-processing side, since modern ASR handles clean single-speaker audio well enough that human time goes to error correction, not transcription. But Rev routes little post-editing work to freelancers, so this row wins on rate and loses on availability. Row 4, competing platforms under the same formula: according to GoTranscript's advertised rates, the maximum is about $0.60 per audio minute; according to TranscribeMe's published rates, the quote is roughly $15–$22 per audio hour. Note the unit bait — an audio-hour quote reads larger than a per-minute quote — yet normalized through one formula, both collapse into the same $6–$12 net band once realistic ratios and contractor tax are applied. Switching platforms does not escape the structural math.

| Evidence item | Source | Figure | Role in the claim test |
| --- | --- | --- | --- |
| Transcriptionist rate | Rev careers page | $0.30–$1.10 per audio minute | Numerator input |
| Captioner rate | Rev careers page | $0.54–$1.10 per video minute | Higher-floor media tier |
| Federal floor | U.S. Department of Labor | $7.25/hour | Default comparator |
| California floor | Dept. of Industrial Relations | $16.90/hour, eff. Jan 1, 2026 | High comparator |
| Washington floor | WA L&I | ~$17/hour, inflation-indexed | Indexed comparator |
| Measured earnings | Glassdoor & Indeed self-reviews | $5–$10 per worked hour | Reality check on the card |
| Completion ratio | AHDI training guidance | 4:1, worsening to 6:1+ | Minutes-to-hours converter |

The explicit winner, within Rev: the captioning track at a sustained $0.80+ per video minute with a personal ratio under 5:1 — at exactly $0.80 and 5:1 the formula returns about $8.13 net, clearing the transcription track's best case with room to spare, and it beats that track at every point of its range. Overall, the winner is any assignment on any platform whose computed net effective hourly meets or exceeds your state's minimum wage. Every other cell loses.

![The Evidence — Rev.com Pay](https://static.mm-ais.com/article-images-pixabay/rev-com-pay-what-up-to-1-minute-actually-18253d02.jpg)

## Track Math

Calibrate before you claim: log media minutes and wall-clock minutes across your next batch of files and compute your personal ratio — the bands above are population-level, and your ratio is the only input you control. Then apply the rule mechanically: claim a file only if (media minutes × rate ÷ your expected hours) × 0.847 meets your state's floor; decline otherwise. On real-world audio — overlapping speakers, far-field microphones — expect your ratio to sit at the unfavorable end of each band, because the bottleneck is disambiguation, not typing speed.

Every threshold in this guide is a snapshot of a moving system. The formula — offered rate times media minutes, divided by your completion hours, times the 0.847 tax factor established above — behaves like a point estimate, and point estimates fail at the edges. Before trusting any row in the preceding tables, understand where the inputs wobble.

**Limitations of the evidence.** The triangulation described earlier — Rev's published card, statutory floors, worker-reported logs — proves less than it appears to. Rev publishes a range, not a distribution: nothing discloses what share of files sits near either endpoint, or how difficulty tracks pay. Completion hours arrive self-reported, which selects for memorable extremes while the routine middle goes unlogged, and survivorship bias is severe — transcribers who quit over pay stop posting numbers. The floors themselves drift: according to Livemint's market coverage, traders spent 2026 reading Fed Chair Kevin Warsh's commentary for cues on the policy path into the fall data cycle, and indexed state wages respond to that same macro weather, so a threshold that clears today can sit below a revised floor within a couple of quarters. None of this overturns the arithmetic; it means the inputs carry error bars the formula silently ignores.

| Track | Quoted rate (native unit) | Completion ratio | Net effective hourly | Verdict |
| --- | --- | --- | --- | --- |
| Rev transcription | About $0.60 per audio minute (observed midpoint, Rev's 2026 rate card) | 4:1–6:1 | $5.08–$7.62 | Decline unless your state's floor sits below this |
| Rev captioning | $0.54–$1.10 per video minute (Rev's 2026 rate card) | 5:1–7:1 | Roughly $7–$11 | Winner inside Rev at $0.80+/minute sustained with a personal ratio under 5:1 |
| ASR post-editing | Machine draft supplied; edit-only work | 1.5:1–2:1 | $12–$20 | Best rate on the board; take it when offered — scarce on Rev |
| GoTranscript | Up to about $0.60 per audio minute (advertised maximum) | 3:1–6:1 | Inside the $6–$12 band | No structural escape |
| TranscribeMe | Roughly $15–$22 per audio hour (published quote) | 3:1–6:1 | Inside the $6–$12 band | No structural escape |

**Variance across cases** is the sharper threat. Media minutes measure duration, not difficulty, and difficulty is where the money hides. In speech-processing terms, the variables that wreck a completion estimate — speaker count, overlap fraction, channel quality, accent mismatch against your ear — appear nowhere in the file listing:

**When the rule breaks — or bends.** Batching first: the formula prices files one at a time, but a fifth short file from a project whose template, voices, and style sheet you already hold inherits sunk setup, so standalone math understates it. Second, the tax comparison is deliberately conservative — employees pay payroll taxes too, so testing after-tax contractor earnings against a pre-tax employee floor double-counts slightly, which argues for declining borderline files, never claiming them. Third, learning curves cut the other way: a first file for a recurring client overstates every future one, so re-run the numbers per project instead of freezing an old estimate. Fourth, the old pitch deserves burial from a new angle: the "$66/hour" claim assumed a human keeps pace with playback, and even a flawless two-to-one sprint on the best-paid class lands around half the advertised figure — reachable only on pristine audio that rarely reaches the open board. Your best file ever is not your expected value.

![Track Math — Rev.com Pay](https://static.mm-ais.com/article-images-pixabay/rev-com-pay-what-up-to-1-minute-actually-87448556.jpg)

## What the Data Doesn't Tell You

The skill to take from this: stratify your own log. Tag each completed file by speaker count and capture quality; once a stratum holds enough runs to show its spread, replace the denominator's guess with that stratum's slow-day median, not your average. Run the formula on worst plausible hours before claiming anything borderline. If a file clears your state's floor only on your best day, it is a decline wearing a tempting sticker.

Start with the document itself. According to the web-search results compiled for this guide, neither endpoint of Rev's published per-minute band is independently corroborated anywhere — both figures trace solely to the company's own headline — and no publicly searchable page breaks the pricing out by project type (transcription versus captions versus subtitles) or lists payout thresholds. The first thing the rate card hides is its own evidence. If you believe the posted band describes what a diligent typist reliably earns, you're reading a leaderboard as a distribution.

The band also describes survivors. Its top endpoint reflects premium files routed mainly to top-graded freelancers, while the median working file — as the midpoint math above showed — sits far closer to the band's floor. Review pools on Glassdoor and Indeed self-select twice over: people post when earnings still justify the bother, and departed low earners go silent. The visible sample is conditioned on continued participation, which inflates every perceived average drawn from it.

| Audio condition | Effect on completion hours | Effect on the claim test |
| --- | --- | --- |
| Single-speaker dictation, clean capture | Sits near the fast end of your personal band | Only class where a modest rate reliably survives |
| Two-speaker interview, good levels | Moderate; turn-taking stays predictable | Borderline files hinge on your logged median |
| Panel or focus group, overlapping talkers | Diarization burden stretches hours toward the 6:1 end of the envelope above | Kills most borderline claims |
| Heavy noise or unfamiliar accents | QA replays multiply; pace collapses | Decline unless the rate visibly overpays the pain |
| Verbatim plus timestamps | Formatting overhead stacks onto listening load | Recompute first; the sticker rate lies |

Completion time is governed by acoustics, not typing speed — the variable I would weight most heavily. Overlapping speech and crosstalk defeat turn-taking segmentation, forcing repeated replay passes; far-field room audio buries consonants in reverberation. Worst are multi-speaker files requiring speaker attribution: diarization errors cascade, because one misassigned turn forces a relisten and a manual re-label of everything downstream. Such files can hit 8:1 ratios even for experts, collapsing effective hourly beneath every floor tabulated above — and Rev's board exposes none of these properties before you claim.

Queue supply is unpriced. Minutes spent refreshing the job board earn $0 and appear in no per-minute statistic, yet they are real worked time. Annualized income therefore depends on daily file volume the worker does not control: two freelancers at identical rates can diverge roughly 2x in yearly earnings purely from queue luck.

## What the Rate Card Hides

Rework is free labor. Customer revision requests and internal quality-control regrades consume additional hours at zero marginal pay, and neither Rev's published ranges nor platform self-reports deduct this time. Every observed effective hourly — including any computed from forum posts — is biased upward until revision hours enter the denominator.

The wage comparison also omits the benefits gap. A W-2 minimum-wage job carries employer-paid payroll contributions and often sick leave or scheduling protections; contractor classification carries none. The 0.847 tax factor defined earlier prices the tax obligation only. At equal nominal hourly rates, the contractor remains materially worse off than the headline comparison suggests — the floors in this guide are generous to Rev, not harsh.

Finally, the 2026 assumption itself is soft. Rev has published no commitment to raise per-minute rates in response to ASR-assisted workflows, so every 2026 projection here extrapolates the 2025 rate card. A unilateral schedule change would invalidate the thresholds computed throughout this guide overnight.

Your new pre-claim habit: classify the audio before accepting — speaker count, room quality, revision likelihood — then run the canonical test with your worst plausible ratio for that acoustic class, not your average. If the adjusted figure clears your state's floor, claim the file; otherwise decline it.

Run the guide's formula on one actual file and the abstractions collapse into a hire-or-pass decision. The specimen: a 45-minute, two-speaker research interview pulled off Rev's freelance board in 2026 at $0.75 per audio minute — mid-card pricing for conversation-grade material. One speaker wears a near-field lapel mic; the other sits across the table on a far-field room mic, with moderate crosstalk where their turns collide. Gross pay: 45 × $0.75 = $33.75. From a diarization standpoint, this is close to the worst profile the board routinely serves: the far-field channel carries reverb and a poor signal-to-noise ratio, and every overlap forces manual re-segmentation of who said what.

Now invert the question: what would this exact file have to pay? At 6:1, each audio minute consumes six worked minutes, so effective hourly is simply ten times the per-minute rate. Matching the federal floor requires $0.86+ per audio minute gross; matching California requires about $2.00 per audio minute — nearly double the ceiling of Rev's published band. No rate the company actually posts makes this recording compliant in a high-wage state. The ratio locks in the outcome before the rate card is ever consulted.

The transferable skill: audition 60 seconds from the middle of the preview — the worst stretch, never the polished intro — and assign a ratio tier before claiming. Clean near-field single-speaker audio earns a 3:1 assumption; mixed channels, room mic, or audible crosstalk gets 6:1 until proven otherwise. Then run the rule mechanically: (offered rate × media minutes ÷ expected hours) × 0.847 against your own state's 2026 floor. On this file, the arithmetic returns its verdict before the preview finishes buffering — and the correct click is decline.

| What the card omits | Mechanism | Impact on net effective hourly | Counter-move |
| --- | --- | --- | --- |
| Premium gating | Top-band files go mainly to top-graded freelancers; the median claimant sees low-band work | Perceived averages inflate | Read the ceiling as a leaderboard entry, not a plan |
| Acoustics | Crosstalk, far-field rooms, and speaker attribution force replay passes; diarization-heavy files reach 8:1 even for experts | Falls beneath every tabulated floor | Substitute your worst measured ratio for multi-speaker audio |
| Queue supply | Board-waiting pays $0 and appears in no per-minute statistic | Identical-rate freelancers diverge up to 2x yearly | Log claim-to-claim gaps as worked hours |
| Rework | Customer revisions and internal QC regrades pay zero marginal dollars | Observed hourlies bias upward | Add revision hours to the denominator |
| Benefits gap | W-2 employer payroll contributions and sick leave have no contractor equivalent | Equal nominal rates still favor the W-2 side | Price lost benefits into your personal floor |
| Schedule risk | No published commitment to raise rates as ASR-assisted workflows spread; 2026 figures extrapolate the 2025 card | A unilateral change voids every threshold here | Re-check the live card each session |

Four dollars. That is the entire gross payout on a ten-minute file priced at $0.40 per audio minute — before the self-employment tax takes its cut, and before you divide by the hours the file actually consumes. Every losing claim on Rev's 2026 board was made by someone who ran this arithmetic after the work instead of before it. The five rules below compress the guide's formula into a pre-claim checklist you can execute in under a minute.

## Worked Case

**Rule 1 — Pre-compute before claiming.** Multiply media minutes by the offered rate, divide by expected completion hours (use 6:1 as the pessimistic default), then multiply by 0.847 to strip the self-employment tax. Decline unless the result meets your state's minimum wage. The ten-minute, $0.40-per-minute file above yields $4.00 gross at best — and that assumes a physically impossible 1:1 turnaround. This is also where the "fast typist" fantasy dies: even a flawless 2:1 performance on a top-of-band file nets roughly $28/hour after tax, attainable only on pristine audio that rarely reaches the freelance board. Speed is not a rescue variable; the clock runs on source-media minutes.

**Rule 2 — Rank tracks by threshold, not by range.** Prefer captioning files at $0.80 or more per video minute over transcription files below $0.60 per audio minute. Inside that band, stop trusting the advertised endpoints entirely and decide from your own last-ten-files data — your measured completion ratio is the only estimator with predictive power, because it embeds your equipment, your pacing, and your topic familiarity.

**Rule 3 — Maintain a rolling personal ratio.** Log hours worked and audio minutes completed for every job. If your trailing 30-day net effective hourly falls below your state floor, stop accepting that file class until your ratio improves or offered rates rise. Treat this as a control loop, not a one-time audit: the board's mix shifts weekly, and yesterday's acceptable track class decays quietly.

**Rule 4 — Treat difficulty markers as price triggers.** More than two speakers, far-field capture, or a flagged crosstalk warning means demanding the top of the range or skipping the file. Overlapping speech is the standing failure mode of speaker diarization — attribution errors compound with each added voice, and the human correction is manual relabeling, minute by minute. Mid-range rates on 6:1-ratio audio lose to minimum wage in every state.

The full gate, applied in claim order:

| Scenario | Gross rate | Ratio | Net $/hr | Verdict |
| --- | --- | --- | --- | --- |
| Interview as posted (2 speakers, lapel + room mic) | $0.75/audio-min | 6:1 | $6.35 | Decline — below every 2026 US floor |
| Same audio priced to reach the federal floor | $0.86/audio-min | 6:1 | $7.28 | Claimable only where state law equals federal |
| Same audio priced to reach California's floor | ~$2.00/audio-min | 6:1 | $16.94 | Unreachable — above the card's ceiling |
| Re-recorded: single speaker, near-field only | $0.75/audio-min | 3:1 | $12.71 | Claim only in federal-floor states |

The transferable skill: audition 60 seconds from the middle of the preview — the worst stretch, never the polished intro — and assign a ratio tier before claiming. Clean near-field single-speaker audio earns a 3:1 assumption; mixed channels, room mic, or audible crosstalk gets 6:1 until proven otherwise. Then run the rule mechanically: (offered rate × media minutes ÷ expected hours) × 0.847 against your own state's 2026 floor. On this file, the arithmetic returns its verdict before the preview finishes buffering — and the correct click is decline.

## How to Choose Well

Four dollars. That is the entire gross payout on a ten-minute file priced at $0.40 per audio minute — before the self-employment tax takes its cut, and before you divide by the hours the file actually consumes. Every losing claim on Rev's 2026 board was made by someone who ran this arithmetic after the work instead of before it. The five rules below compress the guide's formula into a pre-claim checklist you can execute in under a minute.

**Rule 1 — Pre-compute before claiming.** Multiply media minutes by the offered rate, divide by expected completion hours (use 6:1 as the pessimistic default), then multiply by 0.847 to strip the self-employment tax. Decline unless the result meets your state's minimum wage. The ten-minute, $0.40-per-minute file above yields $4.00 gross at best — and that assumes a physically impossible 1:1 turnaround. This is also where the "fast typist" fantasy dies: even a flawless 2:1 performance on a top-of-band file nets roughly $28/hour after tax, attainable only on pristine audio that rarely reaches the freelance board. Speed is not a rescue variable; the clock runs on source-media minutes.

**Rule 2 — Rank tracks by threshold, not by range.** Prefer captioning files at $0.80 or more per video minute over transcription files below $0.60 per audio minute. Inside that band, stop trusting the advertised endpoints entirely and decide from your own last-ten-files data — your measured completion ratio is the only estimator with predictive power, because it embeds your equipment, your pacing, and your topic familiarity.

**Rule 3 — Maintain a rolling personal ratio.** Log hours worked and audio minutes completed for every job. If your trailing 30-day net effective hourly falls below your state floor, stop accepting that file class until your ratio improves or offered rates rise. Treat this as a control loop, not a one-time audit: the board's mix shifts weekly, and yesterday's acceptable track class decays quietly.

**Rule 4 — Treat difficulty markers as price triggers.** More than two speakers, far-field capture, or a flagged crosstalk warning means demanding the top of the range or skipping the file. Overlapping speech is the standing failure mode of speaker diarization — attribution errors compound with each added voice, and the human correction is manual relabeling, minute by minute. Mid-range rates on 6:1-ratio audio lose to minimum wage in every state.

**Rule 5 — Benchmark against your state, converted to gross.** Divide your state's minimum wage by 0.847 to get the gross pre-tax hourly you must clear — about $19.95/hour against California's $16.90. According to the search record compiled for this guide, no indexed source supplies the other states' 2026 dollar floors, and the figures move yearly; pull your own state's current posted schedule before setting the line. That converted number — not the $7.25 federal figure — is the acceptance threshold for every claim.

The full gate, applied in claim order:

| Gate | Test | Action |
| --- | --- | --- |
| Pre-compute | (minutes × rate ÷ expected hours) × 0.847 vs. state floor | Meets floor: claim; falls short: decline |
| Floor sanity check | 10-min file at $0.40/min = $4.00 gross maximum | Clears no U.S. floor at any ratio: auto-decline |
| Track selection | Captioning at $0.80+/video-min vs. transcription below $0.60/audio-min | Prefer captioning; inside the band, defer to last-ten-files data |
| Rolling audit | Trailing 30-day net effective hourly vs. state floor | Above floor: continue; below: halt that file class |
| Difficulty flags | More than two speakers, far-field audio, or crosstalk flag | Top-of-range offer: consider; mid-range: skip |
| Gross benchmark | State wage ÷ 0.847 (California: $16.90 → about $19.95/hr) | Clear it: claim; miss it: decline |

## What to do next

| Step | Action | Why it matters |
| --- | --- | --- |
| 1 | Open a live listing on Rev's freelance board and read the offered rate as a per-audio-minute price, not a wage — a $1.10 top-band file pays $66.00 per source-media hour, not per worked hour. | Rev's meter bills clients per media minute and locks your gross at claim, so treating the banner rate as an hourly wage overstates pay by the full completion-ratio gap. |
| 2 | Before claiming any file, run the canonical screen: (offered rate × media minutes) ÷ your expected completion hours × 0.847, the multiplier that strips the 15.3% self-employment tax. | Only the after-tax, per-worked-hour result is comparable to real wages; gross-per-audio-hour figures hide the 3:1–6:1 production drag entirely. |
| 3 | Compare that result against your state's minimum wage and decline the file if it falls short — start screening with the hardest-audio listings, where completion hours balloon fastest. | Even the $66.00 ceiling cannot reach California's 2026 minimum-wage floor on anything harder than clean audio, so lower bands fail by wider margins. |
| 4 | Stress-test borderline claims across the realistic 3:1–6:1 spread instead of the 1:1 ratio the banner implies — the $0.30 floor's $18.00 per audio hour holds only on paper, within the $18.00–$66 band. | Realized hourly pay falls by the same factor as the ratio; a file that clears your state floor at 3:1 can sink below it at 6:1. |
| 5 | Re-run the screen annually against inflation — US CPI stood at 3.4% (BLS, July 2026 reading) and PCE at 3.7% (BEA, June 2026 reading) — since Rev's fixed per-minute rates erode unless repriced. | An unchanged per-minute rate loses purchasing power every year, so a file that cleared your state floor last year may fall beneath it now. |
| 6 | When benchmarking pay online, discard any token-price tables quoting $0.96–$3 per 1M tokens — that is MiniMax M3 LLM API output pricing on OpenRouter, not Rev compensation. | Attributing AI-compute costs to Rev inflates expectations; transcription pay is set solely by the per-minute rate card and your own completion ratio. |

## Frequently Asked Questions

**If I claim a 60-minute file at Rev's $0.60 midpoint rate, how much do I actually take home per hour worked?**

At the industry-standard 4:1 completion ratio, the $0.60 midpoint pays $36.00 per audio hour but works out to $9.00 gross and $7.62 net per worked hour after the 15.3% self-employment tax.

**Does Rev's advertised $1.10-per-minute top rate really translate to $66 an hour?**

No — the $1.10 rate buys one minute of audio rather than one minute of labor, and even a flawless 2:1 completion on a top-band file yields $33.00 gross, about $28 net, per worked hour.

**What happens to my effective pay when the audio is crosstalk-heavy?**

Crosstalk-heavy files push transcribers to a 6:1 completion ratio, dropping the midpoint rate to $6.00 gross and $5.08 net per worked hour — below the $7.25 federal minimum wage.

**Do captioners earn more than transcriptionists at Rev?**

Captioners sit on a separate tier at $0.54–$1.10 per video minute, but frame-accurate timing and characters-per-line limits push completion ratios toward 5:1–7:1, netting roughly $7–$11 per hour.

**Is there any way to push my effective hourly above $12 at Rev?**

Post-editing an ASR draft instead of transcribing from a blank page cuts the completion ratio to about 1.5:1–2:1 and lifts effective hourly pay to $12–$20 wherever such volume exists.

**What happens if Rev grades one of my returned files below its quality threshold?**

Every returned file is graded, and scores below Rev's threshold lock a freelancer out of higher-paying postings, making the top of the $0.30–$1.10 band a conditional rather than a standing offer.

## Quick answers

| What does Rev's advertised $0.30–$1.10 per-minute range equal per audio hour, and under what condition? | It equals $18.00–$66 per audio hour, but only at a 1:1 completion ratio. |
| --- | --- |
| What is the midpoint conversion Rev never prints? | $0.60 per audio minute—the center of the advertised range—is $36.00 per audio hour of source media, which lands at $9.00 per worked hour at the industry-standard 4:1 completion ratio. |
| How much does the midpoint pay after self-employment tax, and how does it compare to wage floors? | At the 15.3% self-employment rate, $9.00 becomes $7.62—37 cents above the federal minimum wage and $9.28 below California's 2026 floor. |
| What do listings quoting $0.96–$3 per 1M tokens actually describe? | They describe MiniMax M3 LLM API output pricing on OpenRouter—an off-thesis compute benchmark that should never be attributed to Rev.com. |
| What completion ratios do transcriptionists actually face according to industry training bodies? | AHDI teaches that one audio hour takes roughly four worked hours—a 4:1 ratio—stretching to 6:1 or worse on poor audio. |

Also worth reading: **How to convert your audio and video files into text with total accuracy**: [How to convert your audio](https://transcribeall.io/blog/how-to-convert-your-audio-and-video-files-into-text-with-total-accuracy.php) · **Transform your audio and video files into accurate text transcripts with AI**: [Transform your audio and video](https://transcribeall.io/blog/transform-your-audio-and-video-files-into-accurate-text-transcripts-with-ai.php) · **How to convert your audio and video recordings into text in seconds**: [How to convert your audio](https://transcribeall.io/blog/how-to-convert-your-audio-and-video-recordings-into-text-in-seconds.php)

### Related reading

- [7 AI-Based Video Background Noise Removal Methods That Actually Work in Late 2024](https://transcribeall.io/blog/7_ai_based_video_background_noise_removal_methods_that_actua.php)
- [How Text Difference Analysis Tools Actually Work A Technical Deep-Dive into Diff Algorithms](https://transcribeall.io/blog/how_text_difference_analysis_tools_actually_work_a_technical.php)
- [7 AI Writing Tools That Actually Help Students Take Better Lecture Notes in 2024](https://transcribeall.io/blog/7_ai_writing_tools_that_actually_help_students_take_better_l.php)
- [The Science Behind Statistical Analysis When Weekly Data Reviews Actually Improve Performance](https://transcribeall.io/blog/the_science_behind_statistical_analysis_when_weekly_data_rev.php)
- [7 Time-Saving DAW Mixing Shortcuts That Actually Work in 2024](https://transcribeall.io/blog/7_time_saving_daw_mixing_shortcuts_that_actually_work_in_202.php)
- [Talk to the Machine: AI Transcription That Actually Listens to You](https://transcribeall.io/blog/talk_to_the_machine_ai_transcription_that_actually_listens.php)

### Latest

- [MIT Audit: Diarization Costs Valid; Three-Zone ROI Exposes Risks](https://transcribeall.io/blog/mit-audit-diarization-costs-valid-three-zone-roi-exposes-risks.php)
- [Conformer ASR: Front-Ends & LMs Drive Low-Resource Noise Gains](https://transcribeall.io/blog/conformer-asr-front-ends-lms-drive-low-resource-noise-gains.php)
- [MIT SLP 2026 Diarization: Embedding Drift & Pipeline Architecture](https://transcribeall.io/blog/mit-slp-2026-diarization-embedding-drift-pipeline-architecture.php)

Canonical: https://transcribeall.io/blog/revcom-pay-what-up-to-1minute-actually-nets-per-hour.php
Markdown: https://transcribeall.io/blog/revcom-pay-what-up-to-1minute-actually-nets-per-hour.php/index.md