Skip to content
Röst & Tal· LaunchAvailable

New Swedish AI Model Transcribes One Hour of Audio in One Second

Swedish company Klang has released Pianissimo, an open AI model for Swedish speech-to-text that can transcribe one hour of audio in one second and run locally.

By the Aheadline editorial team·23 sep. 2026·1 min read·Source: Entity-watch: AI SverigeVerifierad signalAI-generated
New Swedish AI Model Transcribes One Hour of Audio in One Second
New Swedish AI Model Transcribes One Hour of Audio in One Second
New Swedish AI Model Transcribes One Hour of Audio in One Second
By · Verktygs- & infrastrukturreporter
Last updated
Vad betyder det för mig?

What happened?

Swedish AI company Klang has released its first open speech-to-text model for Swedish, named Pianissimo. The model is built on Nvidia's open Parakeet architecture and has been optimised for Swedish dialects. According to the company, Pianissimo can process one hour of audio material in one second, enabling real-time transcription with very low latency. The model can run locally and offline on a standard laptop.

Key facts

ModellnamnPianissimo
GrundarkitekturNvidia Parakeet
Prestanda1 timme ljud på 1 sekund
DistributionsplattformarHugging Face, Berget AI

Why it matters

The launch means there is now a fast, open Swedish speech-to-text model available to developers without reliance on foreign cloud services. By releasing Pianissimo freely on Hugging Face and via Berget AI's API, other actors can build upon the technology and integrate Swedish voice recognition into their own systems.

Who is affected?

The primary target group is developers, researchers, and organisations wishing to build their own voice applications or transcribe Swedish speech. As the model is open and can be run locally, it is also relevant for users and companies requiring offline management of audio data.

Impact on the EU

Because Pianissimo is distributed as an open model on Hugging Face and via the Swedish cloud service Berget AI, it is immediately available throughout the EU without restrictions. The model's focus on local processing facilitates compliance with strict European data protection regulations such as GDPR.

What else you should know

The model is based on the Nvidia Parakeet architecture and has been fine-tuned by Klang to handle Swedish speech and various regional dialects. Providing the model on Hugging Face encourages further development and integration into other software.

Frequently asked questions

Quick answers about this story

Vad har hänt?
Svenska bolaget Klang har släppt Pianissimo, en öppen AI-modell för tal-till-text som optimerats för svenska och svenska dialekter.
När hände det?
Lanseringen uppmärksammades den 23 september 2026.
Varför spelar det roll?
Modellen gör det möjligt att transkribera en timmes svenskt ljud på en sekund och kan köras lokalt på en vanlig laptop utan extern molnanslutning.
Vilka berörs av nyheten?
Pianissimo riktar sig till utvecklare och företag som vill bygga rösttjänster på svenska eller transkribera samtal i realtid.
Original source
Entity-watch: AI Sverige·computersweden.se

The link opens in a new window and leads to the publisher's own site.

Verifierad signal

Källan har spårats automatiskt från utgivaren via Aheadlines signalkedja.

AI-verktyg i artikeln

Topics

#Voice#Sverige#AI-modell
[ STAY UP TO DATE ]

Get similar news straight to your inbox

No affiliate linksCancel anytimeGDPR-friendly
[ Frequency ]
[ What do you want to read about? ]

You'll receive updates on 2 topics.

The reader's room

Send in a question or an addition. The newsroom reads everything before it's published and replies when relevant. No AI-generated text – just people.

Sign in to submit a comment or question.

Loading comments…
How this affects you

Read the article through your role

  • Decide whether this affects strategy over 6–12 months or is just noise.
  • Discuss with leadership: do we own the right question or does ownership need to move?
  • Ask: what risk are we taking by NOT acting on this this quarter?

Generated angle — not editorial analysis of "New Swedish AI Model Transcribes One Hour of Audio in One Se"