Realtime voice agents that speak Swiss German.
Build voice assistants and phone bots that understand your callers and answer the way they speak: in Züritüütsch, Bärndütsch, Baaseldütsch or Lozärnerdütsch, in the French of French-speaking Switzerland and in the Italian of Ticino, over a single WebSocket. Speech-to-text and text-to-speech stay available as separate building blocks.
No credit card. Billed per second after that.
import WebSocket from 'ws';
const ws = new WebSocket(
'wss://api.suisse-speech.ch/v1/agent/stream',
{ headers: { 'X-API-Key': process.env.SUISSE_SPEECH_KEY } },
);
ws.on('open', () => ws.send(JSON.stringify({
type: 'session.start',
language: 'de-CH', dialect: 'zurich', voice: 'anna',
instructions: 'Du bist die Rezeption der Garage Muster.',
greeting: 'Grüezi, wie kann ich helfen?',
input: { encoding: 'pcm16', sample_rate_hz: 16000 },
output: { encoding: 'pcm16', sample_rate_hz: 24000,
transport: 'binary' },
})));
// Then stream the caller's audio as binary frames
// and play the audio frames that come back.Three products, one API
The same key, the same language codes and the same error format, whether you listen, speak or let an agent hold the whole conversation.
Voice Agents New
A realtime agent that listens, understands and answers with natural speech over one WebSocket. You give it instructions and tools, it holds the conversation.
- Understands and speaks Swiss German dialects, Swiss French and Ticino Italian
- Your instructions, your knowledge, your rules
- Asks your backend to look something up or book it
- Natural interruptions, telephone audio and live transcripts
Speech-to-Text
Swiss German speech in, Standard German text out, with word timings and confidence. Stream it live or send whole recordings.
- Live transcripts over WebSocket, updated every 250 ms
- 60 languages, the one spoken detected on the fly
- Your own vocabulary: names, products, places
- Long recordings as jobs, up to 200 MB
Text-to-Speech
Write in Standard German and hear real dialect, with its own words, grammar and sound. Twelve voices, from telephone to studio quality.
- Zürich, Bern, Basel and Lucerne German, hand-tuned
- Realtime streaming for voice bots, studio model for prepared audio
- Swiss numbers, dates and CHF amounts spoken correctly
- Telephone-ready: G.711 µ-law and A-law at 8 kHz
Why teams choose Suisse Speech
Real dialect, not an accent
Swiss German output uses dialect wording and grammar. It is not Standard German read with a Swiss accent.
Swiss details handled
Phone numbers grouped the Swiss way, per dialect. Dates, times and CHF amounts in their spoken form.
One conversation, 60 languages
Recognition detects which of your expected languages is spoken. Synthesis answers with the same language code.
Your audio is not stored
Audio, text and transcripts are processed in memory. We keep only content-free metadata for billing.
Free sandbox for your tests
Sandbox keys return realistic test audio and transcripts, never billed. Made for continuous integration.
Honest about limits
Every refusal names its reason: your rate, your plan, your credit or our capacity. Nothing is queued silently.
Built for Swiss voice products
Voice bots and contact centres
Let a voice agent hold the whole conversation, or listen with the realtime transcriber and answer with the realtime voice around your own logic. The caller hears the reply in their language and, if you like, in their dialect.
Call transcription and analytics
Turn recorded calls into searchable Standard German text with word timings, as a batch job that runs well below real time.
IVR prompts and announcements
Produce prompts, greetings and announcements in studio quality, in the dialect of your region. Phone numbers are read out the Swiss way.
Media, e-learning and accessibility
Voice-over for content in Swiss German, French and Italian, and transcripts of spoken Swiss German for subtitles and archives.
Pay per second, start with an hour
Voice agents CHF 0.33 per minute (CHF 19.80 per hour) of connected session time, speech-to-text from CHF 1.35 per hour of audio, text-to-speech from CHF 4.60. No subscription and no minimum per request.
Frequently asked questions
What is a voice agent?
A realtime conversation partner over one WebSocket. You stream the caller’s audio, the agent listens, understands and answers with natural speech, following your instructions and, if you define them, calling functions in your backend. You build the phone bot or voice assistant, we run the listening, understanding and speaking.
Can the API transcribe Swiss German dialect?
Yes. Swiss German speech from any region is transcribed. Because Swiss German has no standard spelling, the transcript is written in Standard German, which is also what search, analytics and language models handle best.
Does the synthesis speak real dialect or Standard German with an accent?
Real dialect. For Swiss German, the text is rendered into the dialect: its wording, grammar and pronunciation. Zürich, Bern, Basel and Lucerne German are hand-tuned. 14 further regions are available on a best-effort basis.
Which languages are supported?
Swiss German, Swiss High German, Swiss French and Swiss Italian first. In total 60 languages for recognition and synthesis, from Arabic and Chinese to Turkish and Ukrainian, and more than 150 with regional varieties and dialects.
How is usage billed?
Per second, with no minimum per request. Speech-to-text and text-to-speech are metered per second of audio, realtime and batch at separate rates. Voice agents are metered per second of connected session time at CHF 0.33 per minute. All rates are listed on the pricing page. You top up prepaid credit, and every new account starts with 60 free minutes.
Do you store my audio or transcripts?
No. Audio, text and transcripts are processed in memory and not stored. Results of batch jobs are kept only until you collect them, for 48 hours at most.
Try it on your own audio
Get 60 free minutes of speech-to-text and text-to-speech. No credit card required.