Privacy policy

Audio data

Generated signals, microphone samples and local comparison files stay on your device by default. Raw captured samples are discarded after analysis unless you explicitly choose to download them.

Local audio normalization

The audio normalizer decodes and processes your selected excerpt in browser memory. Audio, filenames, loudness targets and results are not uploaded or saved to site storage. You may download a new WAV copy and a measurement report containing the excerpt format, selected targets, gain and before-and-after measurements. The report excludes filenames, audio, account identifiers and timestamps. Playback-level matching affects listening only. Downloads remain on your device until you remove them.

Local storage

Saved recipes, technician and reviewer acceptance runs, audio baselines, low-frequency listening observations and comparisons, diagnostic-session observations, and commercial-consent preferences use storage in this browser. They are not account-synced. Acceptance runs can be removed on the recipes page or deliberately exported and imported as JSON. You can delete low-frequency records on the subwoofer test page and a saved diagnosis in the diagnostic workspace. Clearing this site's browser data removes local records.

Analytics and commercial content

WaveLocus does not currently load standalone analytics, advertising providers, or affiliate inventory. Optional embedded videos and external destinations have their own connection and service-activity data as described below. If additional analytics or commercial content are introduced, they require a separate choice where applicable and will not include raw audio, local filenames, continuation secrets, detailed hardware fingerprints, or evidence-pack contents.

External changelog

What’s new links open the full changelog hosted by Headway. WaveLocus does not embed or load a Headway widget. The destination is a fixed public URL with no recording, report or session information appended, and the links suppress the referrer. If you follow the link, Headway handles the connection and any storage on its own website.

Optional tutorial videos

Some guides and the ChatGPT recording review page contain a YouTube privacy-enhanced video player. Opening a page containing a player connects your browser to YouTube, even if you do not play the video. Videos do not start playing automatically. Google receives connection information such as your IP address, browser information and the referring site. Video playback and any player advertising are handled under Google's privacy policy. Privacy-enhanced mode limits personalization but does not prevent communication with Google. Loading a tutorial does not upload your WaveLocus recordings, filenames or local reports, or grant microphone access.

Cross-device continuation

If you choose cross-device handoff, the service temporarily receives the structured test state required for that experiment. Raw audio is not relayed. The one-time continuation expires after ten minutes. The service keeps no general account or permanent diagnostic history.

Optional assistant comparisons

The optional assistant service runs at assistant.wavelocus.com. Connecting your assistant does not give it access to recordings or saved browser tests. Creating a microphone comparison sends a temporary investigation record to Cloudflare. Pairing requires your browser approval. Sending results to the paired assistant requires a separate Share comparison action.

Shared fields are the application category, your browser and application observations, whether the same microphone was used, and the factors you changed. These are user-reported observations, not imported microphone measurements. We do not send recordings, microphone names, local filenames, free-text notes or conversation transcripts through this workflow.

The optional recording-review workflow analyses up to three WAV, MP3, M4A/AAC or FLAC excerpts locally, subject to browser decoding support. When you start it from your assistant, a one-use link connects the browser to that authenticated temporary connection. The link expires after ten minutes and is removed from the address bar when the page opens. Approving the connection shares no measurements. A separate Share report action sends only the displayed per-channel sample levels, timestamps, DC-offset and full-scale indicators, method/reference versions, decoding-format category, intended-use category, editor category and revision context. A batch includes only the recordings you select, identified by numbered slots. Sharing replaces the previous report or batch. No raw audio, filename, transcript or device identifier is included. These values are computed by the browser, not independently verified from audio on the server.

Recording-report metadata includes the report type and schema version, excerpt duration, sample rate, channel count and channel numbers. Measurements include each channel's sample peak and its position, average sample level (RMS), DC offset, full-scale sample count and first full-scale position when present. Original and revised values may be included with your same-passage confirmation and the revision provenance. The assistant also receives interpretations derived from those values, including reference comparisons, a conditional next action, measurement limits and whether a comparison is compatible. Local passage-playback locations, filenames and free-text editing notes are not shared.

If you explicitly share an Analysis v2 report, it also includes decoded frame counts, estimated true peak per channel, gated loudness of the selected excerpt in LUFS, loudness availability status and the number of complete analysis blocks. If you choose a background-only passage, its start and end sample positions, duration, sample rate, format category, unweighted measurement basis and your declaration that it contains only background sound are included. That selection includes each channel's RMS, sample peak, DC offset, full-scale count and measured-or-silent status. Both original and revised selections may be included. The tool does not verify that speech is absent, measure equipment noise floor or assess recording quality. Unavailable values remain unavailable, not zero. These additional fields still require a separate Share report action and do not include audio.

The assistant receives a browser launch link containing a random temporary investigation identifier and a short-lived access ticket so it can open and read the correct approved report. Recording-review responses do not repeat the identifier in a separate field. The identifier routes the read request to the approved report. It is not a name, email address, device identifier or account profile. These connection values are not recording measurements and must not be forwarded. Authentication and storage also use temporary hashed connection identifiers, permissions, expiry times, approval state and operation counters. They are used to enforce access, retention and abuse limits, not to identify recording devices or create an advertising profile. Tool results contain no internal account identifiers, request or trace IDs, wall-clock timestamps, debug objects or logs. Positions within an excerpt are audio-relative measurements, not when you recorded or used the service.

Investigation access expires after 24 hours at most. In the recording workflow, open Manage access in Step 4 and choose Delete report to revoke access and remove its active application record. The microphone-comparison workflow offers Delete comparison. These controls require the original connected browser tab. Scheduled cleanup removes expired records; infrastructure failures can delay cleanup but do not extend access. Refreshing the page loses the browser access secret. Pairing codes and unclaimed recording links expire after ten minutes and assistant access tokens after thirty minutes. Do not forward the launch link or publish it.

Connection approval uses a Secure, HttpOnly consent cookie and a temporary approval record that expire after ten minutes. Authorization codes expire after two minutes and can be used only once. Stored access and browser-secret values are hashed; this temporary authorization is not a permanent WaveLocus account.

Shared results become available to the assistant platform under its own policies. Deleting a WaveLocus comparison cannot remove messages already received by that platform. Cloudflare also operates the hosting and storage infrastructure. Its SQLite storage supports recovery of earlier database states for up to thirty days. Deletion from active storage is not a promise of immediate deletion from provider recovery copies. We do not use recovery to extend expired investigation access.

To limit abuse, the service uses a keyed identifier derived from the connecting IP address and UTC day. It stores that identifier, not the raw address, in temporary hourly creation counters. Global daily counters do not contain comparison contents. These controls are not advertising identifiers. Cloudflare necessarily processes network addresses to serve requests.

Questions

For privacy or product questions, email support@wavelocus.com.