Paxa Labs and ElevenLabs Scribe for multilingual transcription
A transcription API for Thai and other languages next to ElevenLabs Scribe, across vocabulary, speaker turns, transcript style, file length, live turns, and what each option adds to the bill.
Thai calls and meetings mix English into Thai speech, and the transcript usually needs timestamps and speaker turns around the text. Scribe v2 lists more than 90 languages, Thai among them, and bills keyterm prompting and transcript editing on top of its hourly rate. Paxa STT Lite transcribes Thai and other languages as spoken, and its vocabulary, word timestamps, and speaker turns add nothing to the charge. It is a research preview, sold at the published rate while its transcripts and options may change between releases. Run your own recordings through both.
Sources:ElevenLabs speech to text overview(opens in a new tab)ElevenLabs API pricing(opens in a new tab)
Rates and claims last checked October 2026
What differs
| Dimension | Paxa Labs | ElevenLabs Scribe(opens in a new tab) |
|---|---|---|
| Thai and other languages | Thai first, and speech in other languages is transcribed as spoken. A Thai sentence that switches into English needs no setting, and nothing is translated. | More than 90 languages, Thai among them. language_code is optional, and the language is predicted when it is left empty. |
| Vocabulary | Up to 50 terms of 50 characters per request, at no extra charge. | Keyterm prompting takes up to 1,000 terms in batch and 50 in realtime, billed as a surcharge on the transcription. |
| Speaker turns | Numbered speaker turns on recordings under 9 minutes, at no extra charge. | Diarization of up to 32 speakers. Labeling the speakers as agent and customer adds a surcharge. |
| Transcript style | The convention option writes numbers as spoken or written, and the style option keeps or drops fillers. Neither changes the charge. | A no-verbatim flag drops filler words, false starts, and non-speech sounds. Transcript editing, billed as a surcharge, rewrites the finished transcript from an instruction. |
| Recording length | Up to 60 minutes and 25 MiB per request with model paxa-stt-lite-v1-preview, in 7 audio formats. | Audio and video files up to 10 hours, uploaded or fetched from a source URL, with results sent to a webhook on request. |
| Live transcription | Paxa STT Lite Realtime decides where each turn of speech ends and sends one finished transcript per turn, with no setting to tune. | Scribe v2 Realtime commits a segment when the client sends a commit, by default, or after a silence threshold the client sets. |
Sources:ElevenLabs speech to text overview(opens in a new tab)ElevenLabs speech to text API(opens in a new tab)ElevenLabs API pricing(opens in a new tab)ElevenLabs realtime commit strategies(opens in a new tab)
- Paxa STT Lite, per hour of audio
- $0.50
- ElevenLabs Scribe
- $0.22
- Their rate, as a multiple
- 0.44×
ElevenLabs prices Scribe v2 and Scribe v2 Medical at one hourly rate on every plan, with keyterm prompting and transcript editing billed on top. Every rate is per hour of recorded audio, from the vendor's cheapest pay-as-you-go model to its most expensive. Per-minute and per-second rates are converted to that unit.
What it costs at your volume
| Audio, per month | Paxa Labs | ElevenLabs Scribe | Difference |
|---|---|---|---|
| 10 hours | Paxa Labs$5 | ElevenLabs Scribe$2.20 | Difference$2.80 more |
| 100 hours | Paxa Labs$50 | ElevenLabs Scribe$22 | Difference$28 more |
| 1,000 hours | Paxa Labs$500 | ElevenLabs Scribe$220 | Difference$280 more |
| 10,000 hours | Paxa Labs$5,000 | ElevenLabs Scribe$2,200 | Difference$2,800 more |
Every figure is the two published rates multiplied by the volume beside it. Where the other vendor publishes a range, the low end is compared and the range is shown.
When ElevenLabs Scribe is the better choice
Scribe takes audio and video files up to 10 hours, from an upload or a source URL. Its keyterm prompting holds up to 1,000 terms, and its diarization separates up to 32 speakers. Entity detection marks 65 entity types in the transcript. A team already using ElevenLabs keeps transcription on the same account and the same pay-as-you-go billing. Choose us when the recordings are Thai, or Thai with English inside it, and the vocabulary and speaker turns should come inside the hourly rate.
Questions
What does research preview mean for production use?
Paxa STT Lite is sold at the published rate while it is evaluated. Its transcripts and options may change between releases, and the changelog on the product page records each change.
Can we keep ElevenLabs for other languages and send Thai to Paxa?
Yes. Nothing in either API prevents routing recordings by language.
Do you publish accuracy comparisons?
No. We publish no accuracy or speed figures of our own, for our model or anyone else's. Transcribe the same recording on both and read the two transcripts side by side.
How current are the prices here?
They come from each vendor's published list rate, recorded in one place and re-checked on a fixed schedule. The month of the last check is at the top of the page.
Other comparisons
- Best Speech to Text APIs for Each Job
- Paxa Labs and Deepgram for multilingual speech to text$0.26-0.31
- Paxa Labs and AssemblyAI for multilingual speech to text$0.15-0.21
- Paxa Labs and OpenAI for multilingual speech to text$0.27-0.36
- Paxa Labs and Google Cloud Speech-to-Text for multilingual transcription$0.18-0.96
- Paxa Labs and Speechmatics for multilingual speech to text$0.24-0.75
Start with free credits
Sign in with Google, GitHub, or Hugging Face and spend 100 one-time free credits on your own text.
Start free