Loading
Preparing your workspace
Developers choose the product outcome, generate a VOAISE API key, install the SDK or copy the snippet for their stack, and launch voice-to-text, voice-to-voice, duplex voice, all-in-one surfaces, or MCP integrations.
{
"profile_id": "rtprof_v2all_support",
"pipeline": "V2All",
"input_language": "auto",
"output_language": "en",
"temperature": 0.2,
"thinking_level": "minimal",
"billing_mode": "developer_metered"
}Microphone audio becomes transcript events.
Spoken user input receives spoken assistant output.
Natural full-duplex voice with transcript events.
Mic, conversation, and typed text route to the right lane automatically.
Expose profile-scoped VOAISE tools through a remote MCP server.
User action selects the lane: mic for transcript, conversation for voice, typed input for text.
Use a VOAISE-hosted realtime session that keeps low-level connection handling inside the platform.
Generate a profile-scoped MCP server URL for approved tools and safe account-aware actions.
Choose transcribe-only, support read-only, tool approval, or outbound-strict behavior.
Record audio milliseconds, session windows, and tool calls for developer billing.
Download JSON, environment variables, JavaScript, React, Python, and MCP config snippets.
V2All starts audio_to_text or audio_to_audio depending on the button.
V2All stays text_to_text and does not request microphone permission.
V2MCP exposes the remote tool server with approval rules.