Rasa Launches Rasa Voice

Rasa, a conversational artificial intelligence platform provider, has launched Rasa Voice, a voice AI architecture that combines structured logic with multimodal input and skips the speech-to-text layer.

Rasa Voice combines live audio and contextual signals to interpret meaning from the first word, without converting everything to text first. This allows the following:

• Faster handoff to execution by skipping transcription bottlenecks.
• Richer interpretation of intent, emotion, and speech patterns.
• More humanlike interactions by preserving the unique qualities of voice instead of reducing it to text alone.
• Tighter alignment between what was said and what the system does.

Rasa Voice is powered by Rasa's CALM (Conversational AI with Language Models) framework. Rather than relying on prompts to deliver outcomes, CALM translates user input into structured, traceable commands that execute within defined business flows. It interprets user speech with nuance, context, and intent, then connects that to business outcomes with clear execution paths.

"We believe voice is becoming one of the most strategic channels for customer experience, and enterprises need reliable solutions," said Melissa Gordon, CEO of Rasa, in a statement. "What we're previewing today is voice automation grounded in structure, designed for scale, and ready to serve the enterprise with the speed, confidence, and nuance users expect."