Klang AI AB, the Swedish company behind the conversational AI platform Klang, has raised SEK 15 million in new funding, an amount equivalent to about 240 million Japanese yen. The capital will support research, development, and international expansion, with Japan identified as a priority market. The company also announced plans to begin training a Japanese-specific speech recognition model in the near future.
Valuation and Investor Backing
The financing values Klang AI AB at SEK 150 million and was led by Johan Lenander. Emil Sjödin, Daniel Gadd, and existing investors also participated in the round, according to information published alongside the announcement. The transaction provides an early indicator of investor appetite for specialized AI businesses in Europe while supporting the company's product and market expansion.
Funding Priorities and Hiring
The new capital will be allocated to research and development, product improvements, and the recruitment of about 10 people. Klang plans to hire across research, development, and marketing to expand its delivery capacity and accelerate work on its own speech recognition models. The company said these investments will help it serve a wider range of customers and strengthen its internal technical capabilities.
European Expansion Plans
Klang will continue to expand in Europe, with a focus on the DACH region and the Benelux region. The DACH region includes Germany, Austria, and Switzerland, while Benelux covers Belgium, the Netherlands, and Luxembourg. These markets form part of the company's broader international growth strategy alongside its push into Japan and its plans for a local Japanese presence.
Japan as a Strategic Market
Japan is a key market for Klang's next phase of growth, and the company plans to invest in both technology and business operations there. The startup will soon begin training a speech recognition model specialized for Japanese, with a focus on compact design, high accuracy, and efficient processing. Over the longer term, it aims to establish a local Japanese presence to serve customers more directly and support local business needs.
Swedish Model Performance
Klang will build on the experience gained from its Swedish model, Klang Pianissimo, which is based on NVIDIA Parakeet v3. The model contains approximately 600 million parameters, and its weights were publicly released on September 23, 2026. In company testing, one NVIDIA A100 GPU processed approximately 41 hours of Swedish audio in one minute through batch processing, though the company noted that Japanese model performance will be validated separately.
Data Quality and Privacy
According to Klang, data quality was a central development priority alongside volume. The team reviewed publicly available speech and text pairings and checked for un-transcribed utterances, removing unsuitable data before training. This process produced more than 50,000 hours of Swedish speech data, and the company said customer conversation data is never used to train AI models.
Product Availability
Klang offers transcription, summarization, and search and analysis tools for meetings and interviews, converting spoken interactions into structured information for customer service, sales, and operational workflows. The service was developed in Sweden and launched its Japanese version in August 2026. It is available via web, iOS, and Android apps, with payment support in Japanese yen and plans for deeper local engagement.
With this funding, Klang AI AB is advancing both international growth and language-specific AI development. Its planned Japanese speech recognition model reflects a broader strategy to adapt technology to local market needs and improve processing efficiency. If successful, the initiative could strengthen the company's position in Japan and reinforce its standing in Europe's specialized AI sector.