KugelAudio: Kugel is an on-prem enterprise TTS supporting 25+ languages

Kugel is an on-prem enterprise TTS supporting 25+ languages

Kugel is an enterprise-ready TTS model, available on-prem, with a focus on 25+ languages and low latency.

Active Founders

Kajo Kratzenstein

Founder

Founder at KugelAudio (YC P26). Previously CTO at Sagemode, scaling an in-game AI coach to 1,000+ users with deep League of Legends integration. Built production aneurysm detection model at Floy GmbH. MA and BSc in Computer Science from Hasso Plattner Institute.

Kajo Kratzenstein

Founder

Founder at KugelAudio (YC P26). Previously CTO at Sagemode, scaling an in-game AI coach to 1,000+ users with deep League of Legends integration. Built production aneurysm detection model at Floy GmbH. MA and BSc in Computer Science from Hasso Plattner Institute.

Viktor Presber

Co-Founder

Co-Founder of KugelAudio, building foundational Text-To-Speech models for customer support. Started founding ventures at 17 with a local video production company. Later co-founded full-house.io, scaling to $300K ARR in six months with enterprise clients like Siemens Energy, Hitachi Rail, and Barilla. Previously co-founded Sagemode, securing partnerships with two major esports teams.

Viktor Presber

Co-Founder

Co-Founder of KugelAudio, building foundational Text-To-Speech models for customer support. Started founding ventures at 17 with a local video production company. Later co-founded full-house.io, scaling to $300K ARR in six months with enterprise clients like Siemens Energy, Hitachi Rail, and Barilla. Previously co-founded Sagemode, securing partnerships with two major esports teams.

Company Launches

KugelAudio (YC P26) is a multilingual voice AI model you can run in your own cluster.

See original launch post

TL;DR: KugelAudio is a multilingual voice AI model that generates natural speech in 30+ languages and can run fully on-prem in your own Kubernetes cluster.

https://youtu.be/R4HqJ8Ai-z4

Launch Day Special 🚀
Book a demo and mention our launch: we’ll give you 20% off.

Problem

Current TTS models are English-first and cloud-hosted. Kugel speaks 30+ languages and dialects natively like Bavarian German, Northern Irish English, or Yemeni Arabic even on edge cases like email addresses, phone numbers, or product names.

Solution
Drop our model into your Kubernetes cluster and use voice AI on patient records, financial transactions, or internal customer data without sending anything to a third party.

A few more things

60ms latency. One of the fastest TTS models on the market.
IPA notation support across 30+ languages.
Voice cloning from a few seconds of audio.

Meet the team

We’re a team of two from Germany and Poland.

uploaded image

Kajo spent the last 2 years researching TTS at Hasso-Plattner-Institut and built the model. Viktor runs commercial and brings the enterprise sales experience. We kept hearing the same thing from enterprise voice buyers: providers mispronounce their product names and their legal team won't sign the DPA anyway. So we built the thing they actually wanted.

🤝 Our Ask

If you deal with mispronunciations or sensitive voice data:

- Book a demo

Thank you! 🙏

Kajo and Viktor