STT language: Sume language_code hint vs the 60 languages MAI lists
Sume STT takes an optional language_code and auto-detects when omitted. A tracker lists MAI-Transcribe-2-Streaming at 60 languages; Sume's docs state no count.
What each side states
Sume's docs say language_code, for example en or ko, is an optional STT hint, and that omitting it means auto-detect. They do not publish a language count. A release tracker lists MAI-Transcribe-2-Streaming as 60 languages with continuous language detection. Sume does not list MAI-Transcribe-2, so this is a facts comparison, not an offer to run it through Sume.
| Item | Sume STT 1.0 | MAI-Transcribe-2-Streaming (tracker) |
|---|---|---|
| Languages | No count in Sume docs | 60 |
| Language control | Optional language_code hint; auto-detect if omitted | Continuous language detection |
| Price | $0.01 per audio minute | $0.54 per hour, introductory |
When to send the hint
Auto-detect is fine for single-language clips. Send language_code when the clip opens with music or silence, mixes languages, or when you caption Korean: for the caption job the language field only tells speech-to-text which language to expect, and it does not choose the typeface.
Pick the hint per source, not per account.
Checking your own audio
Run a handful of your real clips with and without the hint, compare the transcripts, and decide per source. A hint costs nothing extra; the rate is per audio minute either way.
For a 10-hour archive at $0.01 per audio minute, that is 600 minutes, so $6.00 of Sume STT regardless of the hint. Vendor prices come from a third-party release tracker that lists them, not from a Microsoft price page, so confirm them on Microsoft's own pricing before you commit.
Related surfaces
The standalone transcribe route is POST /v1/stt-1.0/transcribe, noted in Timeline audio; the per-minute rate is documented in Video inspect.
Edge cases
Short clips give auto-detect little to work with. A two-second reply in a name or a loanword can be misdetected, so send the hint for short clips in a known language. Code-switching clips are the harder case; transcribe with the dominant language hint and check the other language's passages by ear.
The tracker's phrase for the Microsoft model is continuous language detection, which is a claim about that product and not something to assume for Sume.
What to test
Build a small test set of clips: one clean single-language clip, one noisy clip, one with a song intro and one mixed-language clip. Run each with and without language_code and compare word timing, spelling of names and any dropped sentences. Cost at $0.01 per audio minute is small enough that a 20-clip test of 3 minutes each is 60 minutes, $0.60.
Sources
More in Comparisons
- Suno Premier credits cost 25% less than Pro: and the Sume math
Suno Pro is $8 for 2,500 credits and Premier $24 for 10,000, so a credit is $0.0032 vs $0.0024. At $0.125 a track, Sume breaks even at 64 or 192 tracks.
- Swapping a face in a 12-second clip: manual steps vs one request
What changing a face in a short clip involves when you do it yourself, and what Sume's beta face-swap request replaces. Fields, limits and the cost ceiling.
- Synthesia's 500 signup credits: 5 minutes of video or 1 survey reply
Synthesia lists 500 signup credits: about 5 minutes of video, 6 of lip-sync dubbing or 1 survey response at its listed rates. The same time on Sume in dollars.
- Synthesia's 60 free Interactive Avatar minutes: what they cover
Every Synthesia plan now includes 60 one-time Interactive Avatar minutes at signup. What that buys, what it does not, and the rendered-clip route on Sume.
Written by Sume