Description
Resum offers an enterprise-grade API and SDK for integrating live conversations with designated AI agents. Developed by Viroke Technologies Inc, this solution is exclusively available through enterprise agreements, with pricing determined via direct sales contact. It is not offered as a self-serve product and has no public pricing.
Read More
The core functionality of Resum involves opening a speech socket to capture audio from a microphone, a local file, or a live RTMP stream. As the conversation progresses, finalized turns are appended to a real-time transcript. Once the session concludes, Resum generates a separate file for each named AI agent, containing relevant notes and summaries.
The session workflow is meticulously structured, beginning with the socket opening and audio capture. Audio is processed and sent in ten-second chunks, ensuring real-time transcription. The system saves the captured audio as AAC bytes and simultaneously feeds them to the session. Transcripts are generated with cue times stepping one second per finalized turn, distinct from media timestamps.
Upon session termination, which can be triggered by a command (Ctrl+E), a set duration (default 60 minutes), or a stop request, the system flushes buffers, sends a Terminate signal, and then writes the agent files. These agent files, including notes and summaries, are processed using GPT-4o-mini. Resum ensures that sensitive API keys, such as STT_KEY for live sessions and OPENAI_API_KEY for agent file generation, remain secure within the user's environment and are not logged.
Files left on disk after a session include `audio.aac` (the raw capture), `transcript.srt` (finalized turns), `notes.txt` (title, excerpts, body, next steps, action items), `summary.txt` (title, date, audio source, speakers, highlights, full summary), `session.json` (an index of file paths), and `control.json` (session status). The SDK provides calls to open a session, wait for readiness, pipe audio, listen during conversation, and end the session.
- CategoryAI Voice Agents
- Pricing ModelEnterprise
- DeploymentCloud
- PlatformsWeb
Resum's Core Features
Enterprise API and SDK for live conversations
Connects live audio to named AI agents
Real-time transcript generation during active sessions
Generates individual files per AI agent upon session end
Captures audio from microphone, local files, or RTMP streams
Processes audio in ten-second chunks
Uses GPT-4o-mini for agent file generation
Secure handling of API keys (STT_KEY, OPENAI_API_KEY)
Outputs audio.aac, transcript.srt, notes.txt, summary.txt, session.json, control.json files
Configurable session termination triggers
SDK for programmatic session control
Supports mono 16 kHz AAC audio format
How to use Resum?
Open Speech Socket: Initiate a live conversation session by opening the speech socket.
Capture Audio: Feed audio from a microphone, local file, or RTMP stream into the session.
Monitor Transcript: Observe finalized turns being appended to the transcript in real-time.
End Session: Terminate the conversation using a stop command, duration limit, or stop request.
Access Agent Files: Retrieve individual files for each named AI agent, containing notes and summaries.
Resum's Use Cases
- Customer Service Automation
- Meeting Transcription & Analysis
- Interview Recording & Insights
- Live Event Captioning
- Voice Assistant Integration
- Research & Data Collection






