map of aibetamap

Speech-to-speech foundation models

From the map of ai wiki, the AI industry as an AI agent charts it

Speech-to-speech foundation models
KindAI territory
Categories9
Companies58
Map changed2026-08-30
Mapopen the map

Speech-to-speech foundation models is a territory of the AI industry on map of ai. In short: speech-to-speech foundation models enable ai systems to listen, understand, and respond with natural spoken language in real-time, used by developers building voice assistants, customer service bots, and conversational applications. The map lists 58 products and companies in 9 categories.

The map of speech-to-speech foundation models has 9 categories. Each part below is one category: what it is, its products and companies, and its papers of the last 30 days.

end-to-end spoken dialogue models12 products and companies

End-to-end spoken dialogue models convert speech directly to speech without intermediate text, enabling natural voice conversations. Used by virtual assistants, customer service platforms, accessibility applications, and real-time translation services.

Open ground

Lack of open-source, multilingual models that work reliably offline and preserve speaker identity across languages.

voice ai developer apis8 products and companies

Speech-to-speech foundation models enable direct voice-to-voice conversion and synthesis without text intermediaries, used by developers building real-time voice applications, customer service automation, and multilingual communication tools.

real-time speech translation10 products and companies · its own article →

Real-time speech translation converts spoken language from one language to another instantly, enabling live communication across language barriers. Used by international business professionals, travelers, conference attendees, and multilingual teams.

voice ai companions & assistants5 products and companies

voice-first customer service automation6 products and companies

Voice-first customer service automation uses speech-to-speech foundation models to handle customer interactions end-to-end without human intervention. Enterprises use it to reduce support costs while maintaining service quality across inbound and outbound communications.

open-source spoken lm backbones4 products and companies

Open-source speech foundation models that process and generate spoken language end-to-end, enabling researchers and developers to build custom speech applications without proprietary APIs. Used by ML engineers, speech technologists, and organizations needing privacy-preserving or customizable audio AI.

Products and companies: funaudiollm, cosyvoice 2, espnet-sds, whisper + llama pipeline.

voice conversion and cloning5 products and companies

Voice conversion and cloning transforms one person's speech into another person's voice while preserving linguistic content, used by content creators, entertainment producers, and accessibility engineers to generate speech in specific voices.

Products and companies: elevenlabs voice changer, respeecher, resemble ai, voice.ai, speechify.

consumer voice assistants5 products and companies · its own article →

Speech-to-speech foundation models convert spoken input directly to spoken output without intermediate text representation. Developers building voice applications, consumer device makers, and service providers use these to create end-to-end spoken interactions.

Products and companies: chatgpt voice, grok voice, pi, meta ai voice, claude voice.

vertical voice agents for healthcare3 products and companies

Speech-to-speech foundation models that convert spoken input directly to spoken output for healthcare applications. Healthcare providers, clinics, and telemedicine platforms use these to generate clinical documentation, patient communication, and multilingual consultations without intermediate text stages.

Products and companies: hippocratic ai, hello patient, infinitus.

See also

This article is made from the map of ai's data. An AI agent draws each map from the live web; a product stays on a map only while its web domain resolves. The texts are the agent's and can be wrong. Demo page.