How to Use TurboScribe’s Paid Plan, Translation and SRT Export
TurboScribe’s paid tier raises the ceiling to 10 uploads a day with no per-file minute cap, and its translation mode can turn a Spanish interview into an English transcript in one pass. This walkthrough covers the settings most people never open, plus a cost comparison against running Whisper yourself.
Why would you upgrade from the free TurboScribe plan?
The free plan’s 30-minute file cap is the wall. Anything longer — a two-hour workshop, a full podcast episode, a 90-minute deposition — needs either manual splitting with ffmpeg or the Unlimited plan.
Splitting works. Run ffmpeg -i input.mp3 -f segment -segment_time 1800 -c copy out%03d.mp3 and upload the chunks. But you lose speaker continuity across chunks and end up with five separate transcript files to stitch. After the third time, most people pay the ten dollars.
Decision rule: if you transcribe more than 90 minutes of audio a week, the Unlimited plan is cheaper than your time spent splitting files.
How do you use translation and multi-language mode?
TurboScribe has a translate toggle that changes the output language independent of the input language.
- Same-language transcription — default. Input Spanish, output Spanish.
- Translation mode — pick a target language in the dropdown. Input Spanish, output English text with original timestamps.
- Auto-detect — leave language on “Detect” and it guesses. Works well for single-language files of a minute or more; fails on short clips and code-switched audio.
Translation quality is Whisper-grade, which means good on European languages and noticeably rougher on idioms, slang, and technical jargon. Portuguese to English held up well in testing; Japanese to English lost honorifics and mixed up subjects more than once.
How do you get usable subtitle files?
SRT export is the reason a lot of people land on this tool in the first place. Two settings change the output significantly:
- Segment length. Default chunks run long, sometimes 12-15 seconds, which is well over the 42-characters-per-line broadcast guideline. Set a shorter max duration if your editor respects cue length.
- Punctuation and casing. On by default in large-v3. Turning it off produces all-lowercase raw text — faster to grep, useless for subtitles.
Then fix the things Whisper always gets wrong: timestamps drift by 200-400ms on files with loud music beds, and cue breaks land mid-sentence. A pass through Subtitle Edit’s “fix common errors” catches most of it in under a minute.
How does the cost compare with running Whisper yourself?
Numbers, using public 2026 rates:
| Approach | Cost for 100 hours of audio | Setup effort |
|---|---|---|
| TurboScribe Unlimited (yearly) | About $120 total, flat | None |
| OpenAI Whisper API | Roughly $36 at $0.006/min for turbo | API key, script, error handling |
| Self-hosted Whisper on a rented GPU | $20-60 in GPU hours, variable | Model download, throughput tuning |
| Self-hosted on your laptop | $0 + electricity | An hour of setup, fans spinning for hours |
The API is cheaper per minute on paper. In practice you pay for it in glue code, retry logic, and file-size chunking at the 25MB request limit. If your volume is under 200 hours a month and you need the transcripts anyway, TurboScribe’s flat pricing is a fair trade.
What do power users actually complain about?
- 10 uploads a day is the real bottleneck, not minutes. A batch of 60 one-minute voice memos takes six days on the Unlimited plan.
- No API on any tier. Batch automation is impossible without browser scripting, and their terms discourage it.
- The in-browser editor is thin. You can correct a word, but there is no find-and-replace across a whole transcript, no glossary, no shared vocabulary for recurring names.
- Re-transcription counts as a new job against your daily upload limit, even for a file you already processed.
If those three limits are dealbreakers, jobs like batch voice memos belong on a Whisper API pipeline instead, and TurboScribe should stay the tool you open for the long, awkward files.
Who is this tier for?
Podcasters with back catalogs, researchers working through interview archives, lawyers and journalists who need timestamped transcripts they can quote, and anyone processing non-English audio they need in English. Small teams sharing one account work too, since the file list is account-wide.
Skip it if you need machine-readable output into a database, real-time captioning, or you are running 500+ hours a month where a self-hosted Whisper Docker image pays for itself in a week.
Quick Summary
- Unlimited plan lifts the 30-minute cap and allows 10 uploads per day; free stays at 3 files/day.
- Translation mode converts between input and output language in a single pass, with original timestamps preserved.
- SRT/VTT export needs a segment-length tweak and a subtitle-editor cleanup pass for anything broadcast-facing.
- The OpenAI Whisper API is cheaper per minute but requires your own chunking under the 25MB request limit.
- Biggest weaknesses: no API, no bulk editing, and a 10-upload-per-day ceiling regardless of plan.
