AI that speaks
Swiss German
natively.
The Swiss German AI toolkit: voices, translator, and voice bots. Built natively for dialect, sounds genuinely authentic. From Zürich to Valais.
The Swiss German AI toolkit: voices, translator, and voice bots. Built natively for dialect, sounds genuinely authentic. From Zürich to Valais.
Enter a text and hear how natural Swiss German sounds. Free, no sign-up required.
The generated output may not be used commercially — for commercial enquiries contact us.
Everything for Swiss German with AI — voices, translation, and voice bots, under one roof.
Type text, get a real dialect voice. MP3, WAV and more, free with an account.
Open the studio →High German in, Swiss German out — as text and as spoken audio.
Translate →Holds conversations in real dialect, following your script, with structured notes.
Learn more →REST API and Python SDK for your own integrations. Batch, multiple formats, clear docs.
Read the docs →Every voice has its own character, its own way of speaking, and its own dialect.
Rational, composed, trustworthy
Authentic, down-to-earth, warm
Authentic, friendly, unmistakable
Excited, lively, enthusiastic
Funny, entertaining, likeable
5 million people speak it every day. Not a single AI system supports our language well. We're changing that — with an open API and free access.
From pronunciation to dialect blending — all in one place.
Every voice is customizable. Fine-tune via API or SDK.
Set mood programmatically — friendly, professional, excited, calm, empathetic, narrative.
Mix dialect characteristics. 80% Zurich + 20% Bern. Generate naturally-sounding regional variations.
Custom dictionaries for brand names, technical terms, and abbreviations. ZHAW, SBB, Migros — all correct.
Swiss German has no standard orthography. The same word can be written ten different ways. Our model processes all of them and produces the correct audio.
Not sure which dialect your text is? The model detects it automatically. Or force a specific one.
Generate dialogues with different voices in a single request. Different speakers, different dialects. Perfect for podcasts, audiobooks, training simulations, and interactive stories.
Custom pronunciation rules for brand names, technical terms, abbreviations, and edge cases.
"ZHAW"→/tset-ha-a-ve/"SBB"→/es-be-be/"Migros"→/mi-grohs/"Coop"→/kohp/"Züri HB"→/tsüri ha-be/Generate up to 10 minutes of continuous speech per request. Articles, reports, audiobook chapters.
Send standard German text and receive authentic Swiss German audio output. The model translates the pronunciation, not just the text.
Provide 30+ minutes of recorded speech and receive your own voice clone in every supported dialect.
Every existing TTS reads standard German with a fake accent. We generate real Swiss German speech.
| Google TTS | Amazon Polly | Azure TTS | ElevenLabs | schwiizerdütsch.com | |
|---|---|---|---|---|---|
| Swiss German | ✗ | ✗ | ✗ | ✗ | Native ✓ |
| Dialect selection | , | , | , | , | Dialects ✓ |
| Swiss prosody | ✗ | ✗ | ✗ | ✗ | Authentic ✓ |
| Free spellings | ✗ | ✗ | ✗ | ✗ | Supported ✓ |
| Swiss provider | ✗ | ✗ | ✗ | ✗ | 🇨🇭 ✓ |
Simple REST API. Python SDK. Batch processing and multiple formats.
import schwiizerduetsch
client = schwiizerduetsch.Client("sk_...")
# Simple generation
audio = client.speak(
text="Grüezi, wie gaats Ihne?",
voice="lena",
emotion="friendly", # coming soon
speed=1.0,
format="mp3"
)
audio.save("greeting.mp3")# Simple generation
curl -X POST https://schwiizerdütsch.com/v1/speak \
-H "Authorization: Bearer sk_..." \
-H "Content-Type: application/json" \
-d '{
"text": "Grüezi, wie gaats Ihne?",
"voice": "lena",
"emotion": "friendly",
"speed": 1.0,
"output_format": "mp3"
}' \
--output greeting.mp3Our own model, real dialects, and an API that just works.
No relabeled German or foreign model. We trained our own speech model from the ground up on Swiss German — that's why it sounds real.
High German or any Swiss spelling — our normalisation automatically maps your text to the correct pronunciation. No rule set to learn.
From Zürich to Valais, trained on real Swiss speakers. No generic accent filters — authentic voices.
Simple REST API with a bearer key and Python SDK. Integrated in minutes, up to 10,000 characters per request.
MP3, WAV, OGG, FLAC and PCM. Speed from 0.5× to 2×. Batch processing for many texts at once.
Your text and audio are processed for synthesis and not stored — never logged, never used for training.
Built in Switzerland, under Swiss law (nDSG). A direct line to the people who built it — we pick up the phone.
Built in part on public Swiss research data — including SwissDial (ETH Zürich, CC BY-NC 4.0).
Yes. The model accepts both Swiss German and standard German as input. When you specify a dialect, it produces authentic Swiss German audio even from standard German input text.
Swiss German has no official orthography. People write the same word in ten different ways. Our model handles all common spelling variants and produces consistent, correct audio, regardless of how the text is written.
Swiss German speakers frequently mix in standard German, English, and French. We handle this through a phoneme dictionary: it lets you define the pronunciation of foreign words, brand names, and technical terms so they come out correctly and in the right Swiss German sound.
Yes in principle — with 30+ minutes of recorded speech, a custom voice clone can be created in every supported dialect. For non-commercial projects, get in touch.
Your text and audio are processed only for synthesis and not stored afterwards. We don't log your content and never use it for training. As a Swiss company, we're subject to the nDSG.