Prepare before use
Allow time for a complete test of audio capture, transcription, and prompts before first use. Confirm that assistance is permitted and provide any required notice or obtain consent.
- Use Chrome or Edge with the required audio capabilities and grant the necessary permissions.
- Check the selected opportunity and preparation materials, then choose the correct audio source.
- Test recognition of the other person's voice and confirm that audio sharing is enabled.
- Check your Credits balance and cost notice. See Credits and pricing for estimates.
Choose an audio source
Computer speakers
The microphone picks up both speakers and cannot reliably distinguish them, so silence-based automatic triggering is unavailable. Use a cue phrase or manual mode: after the question, say the configured cue, such as “Let me think,” or choose “Generate suggested answer.”
The generated suggestion stays locked. Your subsequent speech is recorded as your response and can inform follow-up questions without replacing the current suggestion. “Continue to next question” is unavailable during generation; use it after you finish answering. With headphones, the microphone usually cannot hear the other person, so use shared audio or a virtual audio device instead.
Browser meetings
Select the meeting tab and enable “Share tab audio.” The shared source should contain only the other person's voice, allowing automatic or manual triggering.
Desktop meeting applications
Choose the entire screen in the browser's sharing picker and enable system audio. System-audio capture on macOS requires a compatible Chrome version; the current instructions use Chrome 141 or later as the baseline. If your system or browser offers no audio option, configure a virtual audio device and select it as the input.
The sharing picker may request video permission as well. Greenroom uses only the selected audio and does not analyse, store, or upload shared video. This permission does not mean Greenroom records the screen.
Trigger and correct suggestions
Automatic mode is limited to isolated tab or system audio. Choose fast, standard, or relaxed pause thresholds of 1.0, 1.6, or 2.2 seconds. Use a more relaxed threshold when the other person pauses frequently.
New transcription does not interrupt a suggestion already being generated. Confirmed new questions are queued. If a sentence is split too early, use “Merge into current question.” To correct recognition, choose “Edit question,” review the text, then save and regenerate; this explicit action stops the incorrect request and generates a new suggestion.
Questions, suggested answers, and follow-up suggestions remain available in the history list. If output is empty or unreadable, check the error state and connection first. End the session and reconnect if the problem persists, and contact support if needed.
Second screen and shortcuts
The second screen must be another tab or window in the same browser. A phone or another device cannot receive the session. This feature does not conceal the product or bypass detection. Check your screen-sharing area.
The preparation page provides adjustable shortcuts for generating a suggestion, showing or hiding prompts, changing background opacity, and opening the second screen. Check for conflicts with meeting-application shortcuts before use.
Transcription and data
Hosted transcription sends audio to a speech-recognition provider. Greenroom does not retain the audio. Limited company, product, and technical terms with brief context are sent temporarily to improve recognition; the full resume or job description is not sent for that purpose. When transcription is used to generate a suggestion, the relevant text is sent to a model. Answers or preparation materials you save sync under the workspace rules.
If you use built-in browser speech recognition, the browser provider processes audio under its own rules. See the Privacy Policy and Boundaries and data.