Deepslate Founders
BERLIN, Oct. 8, 2026 | Deepslate, a startup building European speech-to-speech AI models, has raised a €7.7 million seed round (about $8.8 million) led by Munich-based 42CAP. The company announced the round Oct. 1. It plans to spend the money on model training, European language data and sales.
Alstin Capital, existing investor SIVentures and several business angels also took part. Co-founder Paskal Paesler said in the announcement, "For our customers, data sovereignty is not a nice-to-have, it is a prerequisite," per Tech.eu.
What Deepslate Builds
Deepslate's models process spoken audio directly, without first converting it to text. Most voice AI systems chain speech-to-text, a language model and text-to-speech. Paesler told Tech Funding News that each stage "introduces error and results in loss of all the sentiment."
The system has three parts: a speech encoder, a reasoning core built on an open-weights language model adapted for European languages, and a speech decoder. The company is ISO 27001 certified and sells through a self-service platform, an API and self-hosting.
Performance Claims
Deepslate says its response time of 440 milliseconds was the fastest speech-to-speech result in an Artificial Analysis benchmark in September 2026. It also says it posted the best error rate for European languages in the CoVoST2 comparison. Both are company-reported results cited in the funding coverage.
Who Is Using It
Deepslate says insurers, contact centers and platform providers use its technology in production. The company has not named specific customers in the coverage reviewed. Tech Funding News reports a team of 17 and lists OpenAI, Google, Grok, ElevenLabs, Gradium, Deepgram and Synthflow as competitors.
Where the Money Goes
Deepslate says it will expand model training, including a German-language data program covering street names, personal names and dialects. It also plans to cut latency, improve voice quality, grow its sales and marketing teams, and scale production infrastructure across EU data centers.
Julian von Fischer of 42CAP said, "Europe needs its own voice models, and Deepslate is one of the few teams." Andreas Schenk of Alstin Capital said the firm invested because it sees "a globally competitive voice AI model lab."
Frequently Asked Questions
How much did Deepslate raise?
Deepslate raised a €7.7 million seed round, about $8.8 million, led by 42CAP. Alstin Capital, SIVentures and business angels also participated.
What is speech-to-speech AI?
Speech-to-speech AI takes spoken audio in and produces spoken audio out without converting to text in between. Deepslate says this preserves sentiment and cuts response time.
Why does Deepslate focus on Europe?
The company says its customers need data sovereignty, with hosting inside the EU and self-hosting available. It also adapts its models for European languages, starting with German.