"How much does captioning cost?" doesn't have one answer, because it's actually five different questions depending on what you're captioning and why. A 90-second internal Slack video and a 45-minute deposition recording are not the same captioning problem, and pricing them the same way is how budgets get blown or quality gets tanked. Below is what each realistic path actually costs, what you get for it, and where each one falls apart.
Professional human captioning services: $1-3 per minute
This is the industry-standard range for a human transcriber (or a human-reviewed AI-plus-QA pipeline) captioning a video from scratch: standard turnaround, non-technical content, typically 3-5 business days. Rush turnaround (same-day or 24-hour) usually adds a premium on top of the base rate — sometimes 50-100% more. Technical, medical, legal, or heavily accented content costs more too, because it takes a human longer to get right and the error tolerance is lower.
Treat "$1-3/minute" as a typical market range, not a quote you can bank on — actual pricing varies by vendor, language, file format requirements, and how many speakers are in the video. A 10-minute video at $1-3/minute lands somewhere between $10 and $30; a 60-minute webinar lands between $60 and $180. For a single video that's real money, but manageable. For a channel putting out 20 videos a month, it adds up to a recurring line item that a lot of solo creators and small teams can't sustain indefinitely.
What you're paying for: near-verbatim accuracy, correct speaker labels, proper punctuation, and — critically — a human who will actually get "Kubernetes" or a strong regional accent right where software still stumbles. This is the tier that makes sense for legal proceedings, medical content, broadcast delivery requirements, or anything where a wrong word in the captions is a real liability, not just an annoyance.
AI-only captioning software and subscriptions: $0.10-0.30 per minute
Automated captioning tools generate captions algorithmically with no human review, and price accordingly — roughly $0.10-0.30 per minute of processed video, or bundled into a flat monthly SaaS subscription with a minutes-per-month allowance. This is 5-15x cheaper than human captioning, and for a lot of content that's the right trade.
The accuracy ceiling matters here: expect roughly 90-95% accuracy on clear audio, and expect that number to drop with accents, technical jargon, overlapping speakers, music beds, or poor recording quality. 90-95% sounds high until you do the math — a 10-minute video at 150 words per minute has around 1,500 words, so even a "good" 95%-accurate pass leaves 75 words wrong. Most of those are minor (homophone swaps, punctuation), but some will be name misspellings or garbled technical terms that a viewer or a search engine won't forgive.
The honest framing: AI-only captioning is fast and cheap enough to caption everything, but it is not "publish and forget." Anything citation-grade, compliance-grade, or meant to be read verbatim (transcripts quoted in an article, captions burned into a paid course, anything under an ADA/WCAG accessibility obligation) needs a human proofread pass on top — which narrows the actual cost gap between "AI-only" and "AI plus light QA" more than the sticker price suggests.
Freelance marketplace rates: the variable middle
Freelance captioners and transcriptionists (sourced through general freelance marketplaces or transcription-specific platforms) typically price somewhere between the AI-only floor and the full-agency ceiling — often per-audio-minute, sometimes per-project or hourly. Rates vary enormously by the freelancer's experience, specialization, and how much editing/formatting is bundled in versus billed separately.
The tradeoff versus a captioning agency: you can often negotiate directly, build an ongoing relationship with someone who learns your content's vocabulary (useful if you cover a recurring niche with consistent jargon), and sometimes get a better rate for volume. The tradeoff versus AI-only tools: turnaround is slower (you're waiting on one person's schedule, not a queue), and quality is less predictable — you're evaluating an individual, not a vendor's QA process. This lane makes the most sense for creators with steady, moderate volume who've found a specific freelancer worth the relationship, not as a first stop for a one-off video.
In-house/DIY: your own time isn't free
Doing it yourself — or handing it to an employee — feels free because no invoice shows up. It isn't. The real cost is your (or their) hourly rate multiplied by how long manual captioning actually takes, and manual transcription reliably runs 4-6x the video's runtime — a 10-minute video takes 40-60 minutes to transcribe and time-sync by hand. (The full breakdown of where that multiplier comes from, and how it compares to automated extraction, is in manual vs. automatic transcription time.)
Run the math at a plausible hourly rate and it stops looking free fast: at $25/hour, that 10-minute video costs $17-25 in labor; at $50/hour (a marketing manager's fully-loaded rate, say), it's $33-50 — comparable to or more expensive than just paying an AI tool $1-3 for the same video and spending 10 minutes proofreading it instead of 40-60 minutes transcribing it from zero. DIY captioning makes sense when the volume is genuinely low, the content is unusually sensitive (nothing should leave the building), or there's simply no budget line for tools and the time has to absorb it — not as a default "it's free" assumption.
The free path: extracting captions that already exist
Here's the option the other four skip past: if the video you need captions for already has captions or a transcript on the source platform, extracting and reusing them costs nothing beyond a proofread pass. YouTube auto-generates captions for most uploads and many creators upload their own human-written captions; when either exists, a tool like SubExtract's video-captions extractor pulls that existing caption track directly — as plain text, copy-paste, or a timestamped .srt file — in seconds, with Pro accounts able to translate the extracted transcript into another language without re-processing the video.
This is genuinely free, but it is not a captioning solution — it's a reuse solution, and the distinction matters. It only works when:
- The video is on YouTube (or another platform with existing caption/transcript support) and already has a caption track, auto-generated or human-uploaded
- The video is public — private, unlisted, or members-only content isn't exposed through this path
- You're repurposing existing content (embedding a transcript in a blog post, pulling quotes, feeding text into another tool) rather than trying to add captions to a video that has none
If a video has zero existing captions, this path produces nothing to extract — you're back to one of the four paid or time-cost options above. The value here is specifically for anyone sitting on a backlog of already-captioned video who needs the text out of the platform and into a usable format, not for anyone trying to caption fresh, uncaptioned footage.
Cost comparison table
| Method | Typical cost | Turnaround | Accuracy | When it makes sense | |---|---|---|---|---| | Professional human captioning | $1-3/min (higher for rush or technical content) | 3-5 business days standard; rush available at a premium | Near-verbatim | Legal, medical, broadcast, or anything where a wrong word is a real liability | | AI-only captioning software | $0.10-0.30/min or flat monthly subscription | Minutes | ~90-95% on clear audio, lower with accents/jargon/noise | High volume, internal or casual content, anything you'll proofread before publishing | | Freelance marketplace | Variable, usually between AI and agency pricing | Days, depends on the freelancer | Depends on the individual | Steady moderate volume with a trusted freelancer relationship | | In-house/DIY | Your hourly rate × 4-6x runtime | Hours per video | As good as the person doing it | Very low volume, highly sensitive content, or no tooling budget | | Extract existing captions | Free (beyond a proofread pass) | Seconds | As good as the source caption track | The source video already has captions and you're reusing/repurposing them |
Which option actually makes sense for a given video
Start by asking whether the video already has captions on its source platform — if it does, extraction is strictly better than every paid option for that specific video, since there's nothing to gain by paying to recreate text that already exists. If it doesn't, the real decision is about stakes and volume: low-stakes, high-volume content (social clips, internal videos, first-draft transcripts you'll edit anyway) fits AI-only pricing; high-stakes, low-volume content (a legal deposition, a compliance training video, a broadcast deliverable) justifies paying $1-3/minute for a human who won't guess wrong on a critical term. Freelancers are worth cultivating once you have steady, moderate volume and have found someone reliable. DIY only wins the math at genuinely low volume or when the content can't leave your organization at all.
Frequently asked questions
Is AI captioning accurate enough to publish without review? For casual or internal content, often yes. For anything citation-grade, compliance-grade, or meant to be read verbatim, no — budget for a proofread pass even at the higher end of the 90-95% accuracy range, since that still leaves real errors in longer videos.
Why is human captioning so much more expensive than AI? You're paying for a person's time to listen, transcribe, format, and QC the whole video — plus their judgment on names, jargon, and accents that automated tools still get wrong. That labor cost is the entire reason the price sits at $1-3/minute instead of $0.10-0.30.
Can I extract captions from a video that doesn't have any? No. Extraction only retrieves captions or transcripts that already exist on the source platform — it can't generate captions for a video with none. If a video has no existing caption track, you need one of the paid or DIY options above.
Does extracting existing YouTube captions cost anything? No, beyond a proofread pass if you plan to publish the text verbatim. A tool like SubExtract's video-captions extractor pulls the existing caption track (auto-generated or human-uploaded, whichever the video has) as plain text or a timestamped .srt file.
What's the real difference between captions and subtitles, cost-wise? Functionally for pricing purposes, none — most services and tools price per minute regardless of whether you're calling the output "captions" or "subtitles." The terms describe different use cases, not different pricing tiers; see closed captions vs. subtitles for the actual distinction.
Next steps
If you're weighing DIY against paying for captioning, read manual vs. automatic transcription time for the full breakdown of the 4-6x time multiplier this guide references. If you're unsure whether you need "captions" or "subtitles" for your specific use case before you start paying for either, closed captions vs. subtitles clears up the distinction. And if the video you need text from already has captions on YouTube, skip the spend entirely and pull them directly with SubExtract's video-captions extractor.