Home›Maximum transcription file sizes and durations, compared (15 October 2026)

Maximum file sizes and durations, compared

What each service says it accepts per file, in its own words, read on 15 October 2026; and what WhipScribe accepts.

The short answer

WhipScribe accepts one file up to 5 GB or 10 hours without splitting or compressing it. Of 49 other services read on 15 October 2026, 44 publish a per-file limit; the table shows each one next to ours.

Every vendor figure is the vendor's own published statement with its URL and read date in the raw data; nothing is inferred. Limits change: quote with the date.

Per-file limits, side by side

ServiceMax file sizeMax durationFree plan noteSource
WhipScribe5 GB10 hoursA free preview of the opening minutes, then packs from $4our page
Amazon Transcribe · Standard batch transcription2 Gigabytes28,800 SecondsFree tier: '60 minutes per month for 12 months' (https://aws.amazon.com/transcribe/pricing/)documented
Amazon Transcribe · Call Analytics batch jobs500 Megabytes14,400 Secondsdocumented
Amazon Transcribe · Medical batch transcription2 Gigabytes14,400 Secondsdocumented
AssemblyAI · Pre-recorded STT, /v2/transcript (audio_url)5GB10 hoursdocumented
AssemblyAI · Pre-recorded STT, local file via /v2/upload2.2GB10 hoursdocumented
AssemblyAI · Sync STT (short clips)40 MB120 sdocumented
Azure AI Speech · Fast transcription (Standard S0)< 500 MB< 5 hours per fileFree (F0): 'Not applicable' in the quota tabledocumented
Azure AI Speech · Batch transcription (Standard S0)1 GBnot statedFree (F0): batch not availabledocumented
Deepgram · Pre-recorded audio API (/v1/listen)2 GBnot statedNo per-file duration cap found; docs do not distinguish upload vs URL fetch.documented
Descript · Free / Hobbyist / Creator / Business1GB / 10GB / 20GB / 50GB per filenot statedFree: up to 1GB per filedocumented
Fireflies.ai · Pro / Business / Enterprise (web uploads)Audio 500 MB; video 2 GB150 minutesFree: audio 500 MB, video 100 MB, 150 minutes per filedocumented
Fireflies.ai · FreeAudio 500 MB; video 100 MB150 minutesSame 150-minute cap per file; video cap 100 MB on Free vs 2 GB on paid.documented
Google Cloud Speech-to-Text · v1 synchronous Recognize10 MB (local files / inline)~1 Minute60 free minutes per month per account (https://cloud.google.com/speech-to-text/pricing)documented
Google Cloud Speech-to-Text · v1 asynchronous LongRunningRecognizenot stated~480 Minutes60 free minutes per month per accountdocumented
Google Cloud Speech-to-Text · v2 synchronous Recognize10 MB1 minute60 free minutes per month per accountdocumented
Google Cloud Speech-to-Text · v2 BatchRecognizenot stated8 hours60 free minutes per month per accountdocumented
HappyScribe · Bulk upload (all plans)1GB per filenot stated10-minute free trial of AI transcription; free plan meeting recordings capped at 45 minutes per recordingdocumented
HappyScribe · Single-file upload (marketing claim)no file size limit (as written)not stateddocumented
Kapwing · Paid plans (Pro / Business), upload into the editor6 GBnot statedFree plan: pricing table says 2 GB, help centre says 250MB (see contradiction). Free auto-subtitling 'Up to 10 minutes'; paid is a monthly minute allowance, not a per-file cap.documented
Kapwing · Free plan (help centre)250MBnot statedFree: 250MB upload per help centre; auto-subtitling 'Up to 10 minutes' per pricing page (allowance, not stated as a per-file cap)documented
MacWhisper · iOS / iPadOS app, cloud transcription (Groq)25MBnot statedCloud mode onlydocumented
Microsoft Word / 365 · Word / OneNote Transcribe, Microsoft 365 subscriptionnot stated300 minutes of uploaded audio per month (monthly allowance, not a per-file cap)Free accounts not mentioned; feature 'only available in Word for Microsoft 365 on Windows in Commercial Tenants'.documented
Microsoft Word / 365 · Word / OneNote Transcribe, Microsoft Copilot licensenot stated30,000 minutes of uploaded audio per month (monthly allowance)documented
Notta · Paid plans, file upload (without speaker recognition)Audio 1 GB; video 10 GB5 hoursFree: 'Up to 3 minutes per conversation' (pricing page)documented
Notta · With speaker recognition, or bilingual transcriptionAudio 1 GB; video 10 GB3 hoursdocumented
Notta · Free / Pro / Business / Enterprise (pricing)not stated3 minutes (Free); 5 hours (Pro, Business, Enterprise) per recordingFree: up to 3 minutes per conversationdocumented
OpenAI Whisper API · Audio Transcriptions API (whisper-1, gpt-4o-transcribe, gpt-4o-mini-transcribe)25 MBnot statedNo free plan. No per-file duration cap stated; file upload only, no URL fetch.documented
Otter.ai · All plans (file import)5 GBnot statedBasic (free): 3 lifetime file imports; 30-minute cap per conversationdocumented
Otter.ai · Basic (free) / Pro / Business / Enterprisenot stated30 minutes / 90 minutes / 4 hours / 4 hours per conversationBasic: 30 minutes per conversation, 3 lifetime importsdocumented
Rev · Rev Legal (court transcripts)under 20GB (total upload, including ZIP)not stateddocumented
Rev · Rev AI (API / dashboard)5TB (web GUI or media_url); 2GB (multipart form-data)17 hours (API docs)documented
Riverside · Externally uploaded recordings (paid plans)25 GB240 minutes (4 hours) per fileMonthly external upload quota 15 hours on Pro/Grow/Webinar/Mobile Plus, unlimited on Business; Free plan not listed.documented
Riverside · Monthly external upload quota by plannot stated15 hours/month external uploads (Pro, Grow, Webinar, Mobile Plus); Business unlimitedFree plan not listeddocumented
Sonix · All plans (web upload)16GB8 hours (unless account enabled for longer)30 min free trial (pricing page)documented
Temi · Pay-as-you-go2GBnot statedFirst order is a free trial. No per-file duration found on Temi's own pages; third-party claims were not used.documented
tl;dv · All plans (uploads, beta)not stated3 hoursFree plan has a lifetime upload count (3 or 5 files, see contradiction). No byte size stated.documented
Transkriptor · All plansnot statedunlimited (auto-split at 10 hours for English and 16 other major languages, 4 hours for all other languages)Trial accounts can transcribe only one file at a time; free tier is a 90-minute account quota, not a per-file cap.documented
Trint · All plans3GB3 hoursdocumented
TurboScribe · Unlimited (paid)5 GB10 hoursdocumented
TurboScribe · Freenot stated30 minutes3 files per day, 1 file at a timedocumented
TurboScribe · All plans (support FAQ)5 GBnot stateddocumented
Veed · Paid plansno per-file limit (recommends project total under 5 GB)no per-file limit stated (recommends project total around 4-5 hours)Free plan: 1 GB per filedocumented
Veed · Free1 GBnot stated1 GB per file, not per projectdocumented
YouTube · Video upload limit (applies to auto-captioned uploads)256 GB12 hoursUnverified accounts are limited to videos 'up to 15 minutes long'documented
Buzz (open-source) · Desktop app (chidiwilliams/buzz), local transcriptionnot statednot statedFree and open source. README and FAQ state no size or duration limit, and do not say 'no limit' either.not found
MacWhisper · macOS app, local transcription (Free and Pro)not statednot statedNeither the Gumroad page nor macwhisper.com states a size or length limit (testimonials mention a 3-hour conversation and a 2.77GB file, which are not documented limits).not found
Rev · Rev.com transcription (standard orders)5 GBnot statedFree tier: 'English only transcription Analyze up to 5 files at a time'; no per-file cap statedinferred
YouTube · Automatic captionsnot statednot statednot found
Zoom · Cloud recording audio transcriptnot statednot statedRequires 'A Pro, Business, Education, or Enterprise account'; Basic (free) not listed.not found

Questions

What is the maximum file size for transcription?

WhipScribe takes one file up to 5 GB and 10 hours. The table compares that with the limit each vendor publishes in its own documentation, read on 15 October 2026; where a vendor publishes none, the cell says so.

Why do limits matter for long recordings?

A limit below your file means splitting it, compressing it or extracting the audio before you start, and a transcript in pieces afterwards. A limit above it means one upload and one timeline.

How were the figures read?

From each vendor's own help or pricing page, with the exact quote, the URL and the date, stored in the raw data. 44 of 49 vendors publish a figure; the rest are marked not stated.

Raw data: /proof/limits-compared.json (every quote, URL and date). What those limits mean in practice: large files, long recordings. Speed, measured: /proof/turnaround.