Optional voice analysis

Notice version: guest-vocal-v2

With your permission, Table Voice records your feedback conversation and uses AI to prepare a report for the restaurant. If you separately agree to optional voice analysis, we also analyse your words and characteristics of speech to help understand your experience.

Your choice

You can leave the optional analysis box unchecked and still give feedback. Participation is optional. You can finish the conversation at any time, or cancel before sending the recording. To withdraw permission or request deletion after sending, contact the restaurant where you gave feedback and tell them the approximate date and time of your visit.

What is processed

Google Cloud stores recordings privately. Existing AI services prepare the conversation and report. For the optional analysis, TypeSafe AI’s Jev receives limited guest transcript evidence and numerical speech measurements. Jev does not receive the recording, your optional contact fields, or assistant speech as evidence of your experience. Google Cloud’s AI model then uses the same limited evidence and Jev’s results to write a short summary and suggested actions for the restaurant; it does not receive the recording or your contact fields for this step. Avoid including personal information in your spoken feedback.

Processing may take place outside India, including in the United States. Optional analysis is enabled only after the restaurant’s approved processor configuration and applicable data-processing arrangements are confirmed.

Access and retention

Approved restaurant listeners can review recordings for up to 60 days. Analysis results follow the source recording and feedback retention period: they become unavailable when their source is deleted or expires. An additional seven-day restricted recovery copy may remain under the existing deletion policy.

How results are used

The initial pilot stores results for evaluation. It does not change guest ratings, charge extra feedback credits, issue new alerts, or affect staff performance scores. Speech patterns are uncertain supporting evidence; they do not establish hidden emotions, truthfulness, or intoxication.