Minimal data by design.

Yapalyzer accepts URLs for public YouTube videos. Temporary audio is extracted and sent to AssemblyAI for transcription so the analysis can run.

iIndex does not retain audio or transcripts after processing. AssemblyAI and YouTube process data under their own privacy terms.

For videos with multiple voices, AssemblyAI may assign anonymous speaker labels such as Speaker A or Speaker B. Yapalyzer stores only the resulting per-speaker counts and scores, not what each speaker said.

When automatic name suggestions are configured, Yapalyzer sends temporary low-resolution speaking frames, limited nearby transcript excerpts, and video metadata to Google Gemini. Gemini is instructed to use explicit names in lower-thirds, title cards, badges, or introductions. A name from the video title may be suggested at low or medium confidence only when the speaker's own first-person context supports it. Faces, resemblance, and biometrics are prohibited. Suggestions require medium or high confidence before Yapalyzer automatically creates a person profile; low-confidence results remain anonymous.

Detailed anonymous results and processing status may be stored for up to 30 days. These records support individual result pages. A permanent video catalog stores public YouTube identifiers, titles, channel and publication metadata, plus aggregate Yapalyzer scores and filler totals. It never stores the transcript or audio.

For medium- and high-confidence identifications, Yapalyzer permanently stores the display name and aggregate per-video statistics for the person profile. Profiles do not contain transcripts, audio, face recognition data, or biometric identifiers. Temporary video, frames, and transcript excerpts used for identification are deleted after processing.

iIndex may use Google Analytics to understand aggregate site usage. Analytics is enabled only when configured by the site operator and is subject to Google's data practices.

Privacy questions can be sent to hello@iindex.io.