Speech-to-Text

Speech-to-text turns spoken audio into text. It is the model kind behind voice-based input and behind transcribing call recordings or voice memos for further processing, for example summarizing a call with Text Generation.

Model

Model ID	adito-stt
Base model	Whisper large-v3
Hosting	ADITO Cloud infrastructure, Germany
Base URL	https://ai.adito.cloud
Endpoint	/audio/transcriptions
Interface	OpenAI-compatible Audio Transcriptions API
Context window	30 seconds of audio per window; longer files are split automatically

adito-stt is the floating alias; it currently resolves to the pinned codename adito-stt-accordion. Naming, the immutability contract and the deprecation lifecycle work the same way here as for every other model kind, see Model naming and versioning.

Default request parameters

Every parameter you do not send is filled in by the runtime with the value below. These defaults are part of what a pinned codename freezes, so they never change for an existing codename — a new codename is issued instead.

Defaults, not limits

A value you send in the request body always wins over the default. Only the audio window is a fixed property of the model rather than something you can override.

Defaults for adito-stt-accordion

Parameter	Default	Description
language	not set	The spoken language is detected automatically. Send an ISO 639-1 code such as de to skip detection and improve accuracy.
response_format	json	Response shape. Also accepts text, srt, verbose_json, and vtt.
temperature	0	Sampling randomness during decoding. 0 gives the most reproducible transcript.
Context window	30 s	The model processes audio in 30-second windows. Longer recordings are chunked by the runtime, not rejected. Not a request parameter.

What you can build

• Voice-based input — let users speak instead of type, for example in a chat interface.
• Call and voice memo transcription — turn a recorded call or voice note into text and attach it to a CRM record.
• Transcript summarization — feed the transcript into a Text Generation request to summarize or classify the conversation.

Audio size and request timeout

The 20 MB limit on the request body includes the audio file itself, and the 120-second timeout covers the complete transcription. As a rough guide, 20 MB is around 20 minutes of audio at 128 kbit/s. Split longer recordings before uploading them, or encode them as mono at a lower bitrate. See Request limits.

First how-to: transcribe an audio file

note

The code examples on this page are not JDito code. They illustrate general API usage and can be adapted to any language or HTTP client.

curl

curl https://ai.adito.cloud/audio/transcriptions \
  -H "Authorization: Bearer your-api-key" \
  -F file="@call-recording.mp3" \
  -F model="adito-stt"

JavaScript

import fs from 'fs';
import OpenAI from 'openai';

const openai = new OpenAI({
  baseURL: 'https://ai.adito.cloud',
  apiKey: 'your-api-key'
});

async function transcribe() {
  const response = await openai.audio.transcriptions.create({
    file: fs.createReadStream('call-recording.mp3'),
    model: 'adito-stt'
  });
  console.log(response.text);
}

transcribe();

The response contains the transcribed text:

{
  "text": "Customer called about extending their existing subscription."
}

---

See also: Text Generation | AI Models