Berlin-based voice AI startup Deepslate has raised €7.7 million in a seed funding round.
The company develops its own speech-to-speech AI models designed to support European languages while meeting data privacy and sovereignty requirements.
The round was led by Munich-based technology investor 42CAP, with participation from Alstin Capital, existing investor SIVentures, and several business angels.
Traditional voice AI usually works in three steps: it changes speech into text, processes the text with an AI model, and then turns the answer back into speech.
Deepslate takes a different approach. Its end-to-end AI model works directly with audio and produces an audio response. The company says this helps reduce response times while keeping important parts of human speech, such as tone, emphasis, dialect, and speaking style.
Deepslate’s system has three main parts, all developed and trained in-house. The first is a speech encoder that understands audio. The second is a reasoning core built on an open-weights language model that Deepslate further trains for specific languages. The third is a speech decoder that turns the AI’s response into spoken audio. The design also allows Deepslate to replace the language model without having to retrain the whole system.
According to Deepslate, its model achieved a 440-millisecond response time in an independent Artificial Analysis benchmark, making it the fastest speech-to-speech model measured by that benchmark as of September 2026. The company also says it achieved the lowest error rate in the comparison for European languages in the CoVoST2 benchmark. Deepslate is also ISO 27001 certified.
The technology is already being used by insurers, contact centres, and other platforms. Customers can access the model through Deepslate’s self-service platform and API, while larger platform providers and enterprise customers can choose volume plans and self-hosting options.
“For our customers, data sovereignty is not a nice-to-have, it is a prerequisite. And either it can be verified or it is worthless. That is why we disclose where the computing happens, who our subprocessors are and what is still open, “said Paskal Paesler, co-founder of Deepslate.
Deepslate will use the new funding to improve its AI models and expand its European data programme. The company will focus especially on German street names, personal names, and regional dialects. It also plans to make its voice AI faster and improve the quality of its conversations.
The company will also expand its sales and marketing teams and increase its infrastructure in European data centres. This will help Deepslate handle more calls as demand for its technology grows.
