Co-Pilot Studios Voice Mode possibility?

Sharma, Paresh 0 Reputation points
2026-04-02T15:21:57.7966667+00:00

I am an educator trying to create agents for learners to practice different scenarios using voice mode. I was wondering if an agent made through Copilot Studio could be shared with learners so they can actually interact with it using their voice in real time, similar to voice modes available in other LLMs. I am specifically looking for live spoken conversation, not voice to text dictation. The goal is for learners to practice their social skills and verbal communication through spoken interaction with an AI agent, not text based communication. Is this currently possible, on the roadmap, or available through any integration? We are an enterprise Microsoft 365 organization.

Microsoft Copilot | Microsoft 365 Copilot | Development
0 comments

1 answer

Sort by: Most helpful
Most helpful Newest Oldest

  1. AI answer

    2026-04-02T15:22:18.3233333+00:00

    For Copilot Studio agents, real-time spoken interaction is supported today, but with important limitations relative to Microsoft 365 Copilot’s own voice chat.

    1. Voice-enabled Copilot Studio agents (IVR/telephony scenarios)
      • Copilot Studio provides a Voice agent template that creates a voice-enabled agent designed for phone-based, hands-free interaction.
      • This template turns on the Telephony channel by default and enables Speech & DTMF (dual-tone multi-frequency) input, so users can speak naturally or press keys on a phone keypad.
      • Voice-specific capabilities include:
        • Speech input with silence detection and speech unrecognized handling.
        • DTMF-enabled multiple choice options for menu-style interactions.
        • A predefined Main Menu topic and an expanded Conversation Start topic tailored for voice.
        • Test chat mode set to Speech & DTMF so authors can test barge-in and DTMF while seeing the agent’s speech responses as text.
      • These agents are intended to be accessed via telephony (phone calls) rather than the same in-app “voice chat” experience that Microsoft 365 Copilot provides.
    2. Voice chat in Microsoft 365 Copilot vs. Copilot Studio agents
      • Microsoft 365 Copilot itself supports two-way voice chat (live spoken conversation) on the web and in the Microsoft 365 app on mobile and desktop. This is the “LLM-style” voice mode experience.
      • However, voice chats are not currently supported in Copilot agents (Copilot Studio agents) within Microsoft 365 Copilot Chat. The documentation explicitly states that voice chats with Copilot agents are not yet available but are planned for the future.
    3. What this means for learner practice scenarios
      • For now, to get live spoken interaction with a Copilot Studio agent, learners would need to interact via a telephony channel (calling into the voice-enabled agent) rather than using the Microsoft 365 Copilot voice chat UI.
      • The richer “in-app” voice chat experience (like other LLM voice modes) is available for Microsoft 365 Copilot itself, but not yet for custom Copilot Studio agents embedded in that experience.
      • The roadmap indicates ongoing investment in speech and IVR in Copilot Studio, including natural text-to-speech, barge-in control, DTMF, and silence detection, aimed at building intelligent voice-responsive agents, but it does not provide a specific date for full parity with Microsoft 365 Copilot’s voice chat.

    In summary:

    • Yes, real-time spoken interaction with Copilot Studio agents is possible today via voice/telephony (IVR) agents.
    • Direct voice chat with Copilot Studio agents inside Microsoft 365 Copilot Chat is not yet supported, though it is on the roadmap.
    • For an enterprise Microsoft 365 organization, the current practical path is to build a Voice agent in Copilot Studio and expose it through telephony for learners to call and practice spoken interactions.

    References:

    AI-generated content may be incorrect. Read our transparency notes for more information.

    Was this answer helpful?

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.