Block launches local voice AI model for business and consumer use
Block is expanding its artificial intelligence work with a new speech model designed to run on consumer hardware while producing more human-like conversations. The model, called Dessa, reflects the company’s push to lower deployment costs for voice applications as it explores broader use cases beyond existing ordering tools.
Highlights
- Block introduces Dessa, a locally running, full-duplex speech-to-speech AI model optimized for MacBook M4 Max or better devices.
- Dessa generates emotion-rich, naturalistic speech—including laughs and pauses—by modeling human expression directly without relying on text-based voice controls.
- Block limits Dessa's release to safety researchers for early testing, addressing risks of voice agents misleading users before broader business or consumer rollout.
Voice model design and rollout
As reported by TechCrunch, Block says its new research model, Dessa, is a full-duplex speech-to-speech system that can talk and listen at the same time while running locally on devices such as a MacBook M4 Max or better. The company says the model is trained fully on human speech and on a single NVIDIA GPU.Block says many current voice systems sound unnatural because text remains in the pipeline that drives speech generation. It argues that Dessa builds emotion directly into the model, producing laughs, pauses and cries without relying on the same style of voice-tag controls used by some competing systems.
The company’s researchers post several conversation examples showing different voices, cadences and emotional tones. In one example, the bot says it is tired from work and its delivery also sounds tired, underscoring Block’s claim that the system produces more realistic expression.
There is no public access to Dessa at the moment. Block is instead asking safety researchers to test the model, a step that likely focuses on risks such as voice agents misleading people in live conversations.
Our earlier report on ByteDance’s AI safety push described how its Seed research unit set up an Alignment & Preparedness team to expand safety training, automated vulnerability testing, monitoring tools, and readiness for emerging risks as frontier models advance. The move underscored a broader shift toward treating safety and independent evaluation as core requirements for deploying increasingly autonomous systems at scale.
-
Afghanistan
-
Albania
-
Algeria
-
Angola
-
Argentina
-
Armenia
-
Australia
-
Austria
-
Azerbaijan
-
Bahamas
-
Bahrain
-
Bangladesh
-
Belarus
-
Belgium
-
Bolivia
-
Botswana
-
Brazil
-
Brunei
-
Bulgaria
-
Cambodia
-
Cameroon
-
Canada
-
Chile
-
China
-
Colombia
-
Congo
-
Costa Rica
-
Cote d'Ivoire
-
Croatia
-
Cuba
-
Cyprus
-
Czechia
-
DR Congo
-
Denmark
-
Dominican Republic
-
Ecuador
-
Egypt
-
El Salvador
-
Estonia
-
Eswatini
-
Ethiopia
-
Finland
-
France
-
Georgia
-
Germany
-
Ghana
-
Greece
-
Haiti
-
Hong Kong
-
Hungary
-
India
-
Indonesia
-
Iran, Islamic republic
-
Iraq
-
Ireland
-
Israel
-
Italy
-
Jamaica
-
Japan
-
Jordan
-
Kazakhstan
-
Kenya
-
Korea
-
Kuwait
-
Kyrgyzstan
-
Laos
-
Latvia
-
Lebanon
-
Lesotho
-
Libya
-
Lithuania
-
Luxembourg
-
Madagascar
-
Malaysia
-
Malta
-
Mauritius
-
Mexico
-
Moldova
-
Mongolia
-
Montenegro
-
Morocco
-
Mozambique
-
Myanmar
-
Namibia
-
Nepal
-
Netherlands
-
New Zealand
-
Nigeria
-
North Macedonia
-
Norway
-
Oman
-
Pakistan
-
Palestine
-
Panama
-
Papua New Guinea
-
Paraguay
-
Peru
-
Philippines
-
Poland
-
Portugal
-
Puerto Rico
-
Qatar
-
Reunion
-
Romania
-
Rwanda
-
Saudi Arabia
-
Serbia
-
Singapore
-
Slovakia
-
Slovenia
-
Somalia
-
South Africa
-
Spain
-
Sri Lanka
-
Sweden
-
Switzerland
-
Syria
-
Taiwan
-
Tajikistan
-
Tanzania
-
Thailand
-
Trinidad and Tobago
-
Tunisia
-
Turkey
-
UAE
-
Uganda
-
Ukraine
-
United Kingdom
-
United States
-
Uruguay
-
Uzbekistan
-
Venezuela
-
Vietnam
-
Yemen
-
Zambia
-
Zimbabwe
- Forex
- Crypto
