Krio ASR v1
Speech to Text
SLE 1
per 5 seconds
- Billed per 5 seconds, rounded up
- Capped at SLE 500 an hour, SLE 900 for two
- Welcome credits spendable here
- Needs SLE 12 on the balance to start
- Audio from 5 seconds up
- Optional noise isolation, free
Pricing
Transcription is billed by the clock, in credits — 1 credit per 5 seconds on Krio ASR v1, capped at SLE 500 an hour and SLE 900 for two. One credit is SLE 1, every new account starts with 12 of them, and a run that fails is refunded. There are no plans and no monthly fee — you buy credits when you need them.
One credit is SLE 1. Failed runs are refunded.
Krio ASR v1 is billed by the clock at 1 credit per 5 seconds of audio, rounded up. One credit is SLE 1, so a three-minute recording costs SLE 36. Longer audio is capped: an hour costs SLE 500 and two hours SLE 900, however long the clip runs in between.
Yes. Every new account gets 12 welcome credits, which is 1 minutes of Krio ASR v1 transcription. Welcome credits are spendable on v1 only.
v1 is the everyday tier, billed at 1 credit per 5 seconds and capped at SLE 500 an hour. v2 is the higher-accuracy tier: it is billed per second with no hourly cap — SLE 600 an hour — charges a one-time SLE 100 unlock from paid credits, and only takes audio from 3 minutes up. Shorter clips run on v1.
Krio ASR v1 needs at least SLE 12 on the balance before it will start a run. It is a floor on what you hold, not on what you are charged — a five-second clip still costs just 1 credit.
No. A run that errors is refunded in full, so a failed transcription costs nothing.
No. Krio text-to-speech is not metered — generating speech in the playground consumes no credits.
No. There are no plans or recurring charges. Credits are bought in bundles, one credit is SLE 1, and they do not expire.
No. Noise isolation is free on both models — it's an optional toggle that separates speech from background sound before transcribing, at no extra credits. The isolated speech track is kept for playback afterward, and on Krio ASR v2 the removed background is kept too.
Ready to build? The speech-to-text API reference has the endpoint, limits and error codes.