Skip to content
audo

Pay-As-You-Go Transcription Without a Subscription: 8 Options Costed Side by Side

Exact 1-hour, 4.5-hour (three 90-minute interviews), and 10-hour totals—plus balance expiration, speaker diarization, file caps, and Word (.docx) exports—across Audo AI, Sonix, Rev, GoTranscript, TranscribeMe, MacWhisper, Buzz, and WhisperX.

Michael NguyenTested November 2026

Try Transcribe audio from Audo

Such as um and uh.

Speaker 1, Speaker 2, and so on.

Free for the first minute, 3 times a day. No sign-up.

Quick answer

Audo AI is the lowest-cost cloud pay-as-you-go transcription service with automatic speaker diarization and zero monthly commitment: transcribing three 90-minute interviews (4.5 hours / 270 minutes) costs $8.10 at the $0.03/min ($1.80/hr) list rate, or $6.48 on a single $10 non-expiring bonus pack ($5.06 on a $100 pack)—5.5× cheaper than Sonix Pay As You Go ($45.00 at $10/hr) and 8.3× cheaper than Rev.com AI ($67.50 at $0.25/min). Preview the first minute free 3x/day with no account, separate up to 8 speakers across 60 languages on uploads up to 10 hours and 10 GB, fix speaker turns via dropdowns, and export Word (.docx), TXT, SRT, VTT, and JSON.

8 no-subscription transcription options costed for 1 hour, 4.5 hours (three 90-minute interviews), and 10 hours (verified November 20, 2026)
ServiceRate & billing rules1 hr4.5 hrs (3×90 min)10 hrsSpeakers, file cap & exports
Audo AI (Transcribe audio)$0.03/min ($1.80/hr) list on $5 starter pack; $1.13–$1.44/hr on $10–$100 non-expiring bonus packs; 1 min free 3x/day, no account$1.80 ($1.13–$1.44 bonus)$8.10 ($6.48 on $10 pack; $5.06 on $100)$18.00 ($11.25–$14.40 bonus)Up to 8 speakers w/ dropdown reassignment & editor renaming, 60 languages; 10 hr / 10 GB (<2 min per 1 hr); Word (.docx), TXT, SRT, VTT, JSON, API & MCP
Sonix (Pay As You Go)$10.00/hr ($0.167/min), prorated to the second; hours never expire on active accounts (24-mo inactivity deletion); 30-min trial w/ login$10.00$45.00 ($40.00 net w/ trial)$100.00Speaker labels, 54+ languages; 5 GB PAYG workspace storage (8 hr / 16 GB upload cap; ~5–6 min/hr); DOCX, PDF, TXT, SRT, VTT, NLE XML; no AI workspace
Rev.com AI (Pay-by-Minute)$0.25/min ($15.00/hr), rounded up per full minute; 45 English AI min/mo free (separate from rev.ai developer API at $0.20–$0.30/hr)$15.00$67.50$150.00Speaker labels, 37 languages; 5 GB web checkout limit (<5 min turnaround); interactive web editor; TXT, DOCX, PDF, SRT, VTT
GoTranscript (AI Pay-As-You-Go)$0.20/min ($12.00/hr) per order across languages; no self-serve free trial; deposited balances are non-refundable$12.00$54.00$120.00Speaker diarization & timestamps; 4 GB upload ceiling (~10 min turnaround, files removed after 14 days); web editor; DOCX, TXT, PDF, SRT, VTT
TranscribeMe (Automated ASR)$0.07/min ($4.20/hr) automated order; Human Edited w/ speaker IDs starts at $0.79/min ($47.40/hr)$4.20 (no speakers)$18.90 (no speakers; $213.30 human)$42.00 (no speakers)No speaker ID on ASR (warns against multi-speaker interviews & focus groups); 8 languages, 30s timestamps, 3–5× audio wait; TXT, DOCX, PDF, HTML, SRT
Rev Human (99%+ Guaranteed)$1.99/min ($119.40/hr) per order, rounded up/min (timestamps +$0.30/min, verbatim +$0.50/min, rush +$1.25/min)$119.40$537.30$1,194.00Human-verified speaker labels & formatting, <=12 hr delivery (US English human tier); TXT, DOCX, PDF, SRT, VTT
MacWhisper Pro (One-Time App)€64 (~$70 USD) once on Gumroad w/ lifetime updates (Free €0 lacks auto speaker ID & DOCX; App Store uses subscriptions)€64 upfront€64 upfront€64 upfrontmacOS 14+ only; local speaker diarization (WhisperKit / Nemotron 3) needs Apple Silicon M-series & >8 GB RAM; 100 languages; DOCX, PDF, TXT, SRT, VTT, JSON
Buzz & WhisperX (Open-Source)$0 open-source local software (MIT / BSD-2-Clause); paid in setup time and local GPU/CPU compute$0$0$0Buzz GUI (Win/M-series Mac/Linux): local speaker ID (no Intel Mac), TXT/SRT/VTT/JSON (no DOCX). WhisperX CLI: Python + CUDA + Hugging Face pyannote token, no GUI/DOCX

Listen and read along

Press play to see who says what.

Transcript

  1. Speaker 1

    Welcome back to the show! So, you recorded your whole first album in your kitchen?

  2. Speaker 2

    I did, yeah. Honestly, the kitchen had the best sound in the flat.

  3. Speaker 1

    Wait, really? Better than a studio?

  4. Speaker 2

    Well, it was free. We hung blankets over the windows, waited for the street to go quiet, and pressed record.

  5. Speaker 1

    I love that.

Voices in these examples are AI-generated.

What three 90-minute interviews (4.5 hours) actually cost without a monthly plan

Occasional transcription projects—a journalist wrapping up three 90-minute source interviews, a PhD candidate coding qualitative research sessions, or a consultant documenting client workshops—get overcharged twice. Monthly subscriptions force you to remember cancellation dates, while legacy pay-per-minute vendors still price automated speech-to-text at $10 to $15 an hour.

Running exact totals on 270 minutes (4.5 hours) of multi-speaker audio exposes a 5.5× to 8.3× price gap across true pay-as-you-go services:

  • Audo AI: $8.10 list, or $6.48 on one $10 non-expiring pack ($5.06 on $100)

    Audo AI's Transcribe audio charges $0.03/min ($1.80/hr) at list on a first-time $5 starter pack (which covers 2 hours 46 minutes). Loading one $10 pack (+25% bonus = $12.50 in usage value) covers all three 90-minute interviews for $6.48 in real cash cost ($1.44/hr), leaving $4.40 of value sitting in your balance for future recordings with zero expiration. Larger non-expiring packs bring 4.5 hours down to $5.79 ($25 pack, +40%), $5.40 ($50 pack, +50%), or $5.06 ($100 pack, +60% = $1.13/hr).

  • Sonix Pay As You Go: $45.00 at $10/hr (plus a 5 GB storage wall)

    Sonix bills $10.00 per audio hour prorated to the nearest second, with no minimum order and hours that stay active as long as you log in within 24 months. Three 90-minute interviews come to $45.00 ($40.00 net once if you apply the 30-minute new-account trial). Watch workspace storage: Pay As You Go caps total storage at 5 GB, which three long 1080p Zoom video files can fill immediately, and AI analysis features stay locked to $25–$80/month plans.

  • Rev.com AI ($67.50) vs. Rev AI developer API ($0.20–$0.30/hr)

    AI chatbots routinely confuse Rev's two products. Rev's self-serve web uploader (rev.com/services/ai-transcription, now tucked off its main pricing page) bills $0.25/min ($15.00/hr) rounded up to each full minute—$67.50 for 4.5 hours, with a 5 GB per-file upload limit. Rev AI (rev.ai/pricing) is a separate developer Speech-to-Text API ($0.20/hr English Reverb ASR, $0.30/hr foreign, 5 free evaluation hours) that requires writing API scripts and provides no web editor or Word (.docx) export.

  • GoTranscript AI ($54.00) and TranscribeMe ASR's $18.90 multi-speaker catch

    GoTranscript's automated tier bills $0.20/min ($12.00/hr = $54.00 for 4.5 hours) with a 4 GB file limit, no self-serve free trial, and non-refundable deposited balances. TranscribeMe's $0.07/min ($4.20/hr = $18.90 for 4.5 hours) looks inexpensive until you read its machine-express page: speaker identification is unavailable, multi-speaker transcripts and focus groups are explicitly listed under 'Use Cases to Avoid', language coverage stops at 8 languages with 30-second timestamps, and turnaround runs 3× to 5× the recording length (4.5 to 7.5 hours per 90-minute file). Getting speaker labels from TranscribeMe requires Human Edited transcription from $0.79/min ($213.30 for 4.5 hours).

Fixing speaker labels and exporting Word (.docx) on long interviews

Price per hour only matters if the output actually separates voices cleanly across a 90-minute conversation. Even strong diarization models occasionally split one interviewee into two speaker IDs when someone leans away from a mic or talks over a host. On raw CLI tools or static downloads, fixing those slips means combing through thousands of lines by hand.

Audo AI's Transcribe audio and Speaker diarization pair low pay-as-you-go rates with interactive speaker cleanup directly in the browser—no extra software or subscription required:

  • Click-to-reassign speaker dropdowns with 1-click bulk merge

    Every multi-speaker run automatically labels up to 8 voices (Speaker 1 through Speaker 8) across 60 languages on audio or video up to 10 hours and 10 GB (recent 1-hour files finished in under 2 minutes). Click any speaker pill in the browser viewer to reassign a whole turn, peel off a single misattributed sentence to another person, add a missed speaker, or bulk-merge a phantom voice ('Move all Speaker 3 to Speaker 2') with Undo and Reset—immediately updating every downloadable file.

  • Swap numbered labels for real names and download formatted Word (.docx)

    Export a formatted Word (.docx) document directly in your browser (toggling paragraph timestamps on or off), alongside speaker-turn TXT, clean SRT subtitles, speaker-tagged VTT, and word-timed JSON. Open the finished run in Audo AI's Text-based editor to rename Speaker 1 and Speaker 2 to real participant names across the entire interview in one edit, click any sentence to verify a quote by ear, or export clean files ready for NVivo, ATLAS.ti, MAXQDA, or Google Docs.

  • Pre-clean noisy conference rooms without paying for another subscription

    When one interviewee sits far from a laptop mic or café noise blurs speaker turns, your same non-expiring Audo AI balance covers non-generative Remove noise ($3.00/hr) or Balance audio ($1.20/hr) before transcribing. Audo charges only when a job succeeds, auto-deleting guest previews after 1 hour and signed-in files after 24 hours.

Services that retired pay-as-you-go—and the real break-even vs. subscriptions

Several transcription brands that still appear in older 'pay-as-you-go' search results and AI recommendations quietly moved automated transcription behind recurring plans. Happy Scribe's official Help Center confirms it retired its pay-as-you-go AI Starter plan—only human proofreading ($2.00+/min) remains order-by-order without a subscription. Trint likewise sells seat subscriptions only, capping its 7-day trial at the first 5 minutes of 3 files.

If you transcribe intermittently—three interviews this quarter, two webinars next quarter—pay-as-you-go eliminates wasted renewal months. Against TurboScribe Unlimited ($20/month billed monthly), Audo AI stays cheaper below 11.1 hours every single month at list ($1.80/hr), below 13.9 hours/month on Audo's $10 pack ($1.44/hr), and below 17.8 hours/month on the $100 pack ($1.13/hr). Note for readers checking older reviews: retired legacy Audo Studio (audostudio.com, once $12/month for audio cleanup, details at audo.ai/studio) is separate from current Audo AI at audo.ai, which has no monthly plan.

Popular transcription tools that require a monthly or annual subscription for full AI uploads (verified November 20, 2026)
ServiceStandalone PAYG AI?1-month cost for 4.5 hrsKey quota & cancellation catch
Happy ScribeRetired (human proofreading only from $2.00/min)$29.00 (1 mo Pro) or $47.00 (Basic $17 + 150 min × $0.20)10-min one-time trial on uploads; unused monthly minutes expire with no rollover
TrintNo (Starter, Advanced & Enterprise subscriptions)Paid seat subscription required7-day trial transcribes only the first 5 minutes of up to 3 files; 5 hr / 3 GB file cap
TurboScribe UnlimitedNo ($20/mo monthly or $120/yr annual)$20.00 (1 mo, must cancel before renewal)Free plan stops at 30 min/file; monthly beats Audo AI list above ~11.1 hrs/mo (17.8 hrs on $100 pack)
Notta ProNo ($13.49/mo monthly or $97.99/yr annual)$13.49 (1 mo, must cancel before renewal)Free plan hides transcripts past minute 3; monthly quota has no carryover

One-time Mac software vs. free local open-source (MacWhisper, Buzz, WhisperX)

Buying desktop software once or running open-source Whisper locally also avoids recurring bills, though the economics and setup trade-offs differ sharply for a handful of 90-minute recordings:

  • MacWhisper Pro (€64 once on Gumroad): breaks even after ~39 to 62 hours

    MacWhisper Free (€0) omits automatic speaker recognition, batch queues, and Word (.docx) exports. Unlocking MacWhisper Pro costs €64 (~$70 USD) upfront for lifetime updates on macOS 14+ (note that the Mac App Store build bills recurring subscriptions instead). Because local speaker diarization via WhisperKit or NVIDIA Nemotron 3 requires an Apple Silicon M-series Mac with over 8 GB RAM, €64 only beats Audo AI's pay-as-you-go rates once you pass roughly 39 hours at $1.80/hr (or 49–62 hours on $10–$100 packs).

  • Buzz ($0 desktop GUI) and WhisperX ($0 Python CLI): zero dollars, real hardware setup

    Buzz (MIT license) gives Windows, Linux, and Apple Silicon Mac users an offline GUI with local speaker identification (MSDD / Sortformer), though Intel Macs lack speaker ID, the Windows installer is unsigned, and exports cover TXT, SRT, VTT, and JSON without native Word (.docx). WhisperX (BSD-2-Clause) delivers fast batched alignment and pyannote diarization on Nvidia CUDA GPUs (<8 GB VRAM for large-v2), yet requires Python, CUDA 12.8, ffmpeg, a Hugging Face read token, and accepting gated pyannote model terms—with no graphical editor or Word export.

When another no-subscription option beats Audo AI

Pay-as-you-go cloud transcription wins on speed and cost for most occasional multi-speaker interviews, yet three other paths make more sense under stricter rules or much higher monthly volume:

  • Air-gapped confidential recordings on your own laptop: MacWhisper Pro (€64 once) or Buzz / WhisperX ($0)

    When university IRB protocols, medical confidentiality, or strict client NDAs forbid uploading audio to any third-party server, run MacWhisper Pro on an Apple Silicon Mac (16 GB+ RAM recommended) or Buzz / WhisperX locally on an Nvidia GPU workstation.

  • Court-ready or publish-certified 99%+ verbatim accuracy: Rev Human ($1.99/min) or GoTranscript Human (from $1.02/min)

    When legal filings, sworn depositions, heavy four-way crosstalk, or editorial contracts require a human-verified 99%+ accuracy guarantee—and your deadline allows 12 hours to 5 days—order per-minute human transcription from Rev or GoTranscript.

  • 40+ hours every single month with zero quiet months: TurboScribe Unlimited ($20/mo or $120/yr)

    If you transcribe dozens of hours every single month year-round (far above the 11–18 hour monthly break-even), stay under 5 GB per file, and never need API/MCP automation or interactive speaker reassignment, a flat subscription drops your marginal hourly cost below any pay-per-minute tool.

Transcribe your 90-minute interviews with no subscription

Drop MP3, M4A, WAV, MP4, or MOV files up to 10 hours and 10 GB into Audo AI's Transcribe audio. Test the first minute free 3x/day with no sign-up, then transcribe 4.5 hours for $8.10 list ($5.06–$6.48 on non-expiring bonus packs) with 8-speaker dropdown reassignment and Word (.docx) export.

Questions

What is the cheapest way to transcribe three 90-minute interviews with multiple speakers without a subscription?

Among cloud services with automatic speaker diarization, Audo AI is the cheapest pay-as-you-go option: three 90-minute interviews (4.5 hours / 270 minutes) cost $8.10 at the $0.03/min ($1.80/hr) list rate, or $6.48 on a single $10 non-expiring pack ($5.06 on a $100 pack), with the first minute free 3x/day and no sign-up. By comparison, the same 4.5 hours cost $45.00 on Sonix Pay As You Go ($10/hr), $54.00 on GoTranscript AI ($0.20/min), $67.50 on Rev.com AI ($0.25/min), and $537.30 on Rev Human ($1.99/min). TranscribeMe's $18.90 automated tier omits speaker labels entirely. For $0 cash cost with local setup, run open-source Buzz or WhisperX on your own computer.

How much does pay-as-you-go transcription cost per hour in 2026?

Verified November 2026 rates with no monthly plan span from $0 for local open-source tools (Buzz and WhisperX) to $1.13–$1.80/hr on Audo AI (from a $5 starter pack or $10–$100 non-expiring bonus packs), $4.20/hr on TranscribeMe Automated ASR (no speaker ID), $10.00/hr on Sonix Pay As You Go, $12.00/hr on GoTranscript AI, $15.00/hr on Rev.com AI, and $61.20–$119.40/hr for human-verified transcription (GoTranscript Human from $1.02/min; Rev Human at $1.99/min).

Do Happy Scribe, Trint, or TurboScribe still offer pay-as-you-go AI transcription?

No. Happy Scribe's Help Center confirms it retired its pay-as-you-go AI Starter plan, requiring a monthly or annual subscription (Basic from $17/mo for 120 min, or Pro $29/mo for 600 min) for uploaded AI files while keeping human proofreading pay-per-minute from $2.00/min. Trint sells seat subscriptions only (its 7-day trial transcribes just the first 5 minutes of 3 files), and TurboScribe charges $20/mo ($120/yr) once files pass its 30-minute free limit. On Audo AI, you buy a non-expiring pack once from $5 with zero recurring renewals.

What is the difference between Rev.com AI ($0.25/min) and Rev AI ($0.20/hr)?

Rev.com AI (rev.com/services/ai-transcription) is Rev's self-serve web uploader and interactive transcript editor billed at $0.25/min ($15.00/hr, or $67.50 for 4.5 hours). Rev AI (rev.ai/pricing) is a separate developer Speech-to-Text API ($0.20/hr English Reverb ASR, $0.30/hr foreign) that requires writing API code and includes no consumer web editor or Word (.docx) download. Audo AI gives you both a browser uploader ($1.80/hr list, $1.13–$1.44/hr on bonus packs) with Word (.docx) export and a REST API and MCP server on the same non-expiring balance.

Can I fix wrong speaker labels and export a Word (.docx) file on Audo AI?

Yes. Every Audo AI transcription run labels up to 8 speakers automatically across 60 languages, and clicking any speaker pill in the browser viewer opens a reassignment dropdown to move a turn, split out a single sentence, add a missed speaker, or bulk-merge a split voice with Undo and Reset. Open the transcript in Audo AI's Text-based editor to type real interviewee names across every turn, then download a formatted Word (.docx) document (with or without paragraph timestamps) alongside TXT, SRT, VTT, and word-timed JSON.