BlindMixRoom — blind mix feedback, offline

BlindMixRoom prepares a blind listening packet for collaborators. Choose two to four already-aligned WAV versions, select one common excerpt, and ask one neutral listening question. Export a single HTML file that collaborators open in desktop Chrome.

The packet includes its free viewer and neutral randomized version labels. Original filenames, producer labels, the reveal map and original WAV metadata stay out of the listening packet. Recipients press Play, compare versions at the same position, loop if useful, choose a preference or no preference, add optional notes, and download a small response file. They select no audio files and install no application to listen.

Import returned response files in your local creator. Review counts, ignore duplicate response IDs, reject files from another session or files that fail integrity checks, and reveal producer labels locally. Save a private project to reopen your original audio and results later.

The $19 one-time download includes the four-version creator, quickstart, recipient instructions, license notices and four owned synthetic practice WAVs. The free two-version creator gives you the same workflow with two versions. Recipients can use packets from either creator without payment or an account. There is no subscription, activation service, upload server or built-in file transfer service.

AI-assisted development: BlindMixRoom was developed with OpenAI Codex, which generated application code, interface wording, documentation and listing copy. The bundled practice WAVs and viewer-demo audio are deterministic synthetic test signals, rendered locally from waveform-generation code written with AI assistance. They are not human-recorded performances and were not produced by a generative audio model. Screenshots capture the actual application using synthetic audio and scripted QA responses. Listener choices and notes are entered by listeners; BlindMixRoom uses no AI model to listen to audio, rank mixes or produce listener judgments.

Requirements and limits: desktop Chrome; integer PCM WAV at 44.1 or 48 kHz, mono or stereo. Versions need the same sample rate and channel count and must already be aligned. All imported original files together must fit 32 MiB, including metadata; this cap applies before excerpt selection. The common excerpt is 1–60 seconds. Exported packets and saved projects are capped at 64 MiB. Use short aligned source WAVs from your DAW when a full-track batch is too large. Float/compressed WAV and other formats are unsupported.

Integrated loudness matching only attenuates louder excerpts toward the quietest, with explicit bypass. Playback starts on a user action at −18 dB gain, uses short fades, and may resample in the browser. Adjust your device volume too. Keep the producer project private: it contains original audio, names, labels and the reveal map. Send only the listening HTML.

Early-release limitations: automated offline QA used owned synthetic audio in macOS Chrome. Physical listening quality, real producer/recipient usability, actual external file transfer and Windows/Linux hardware remain unvalidated. HTML attachments may be blocked or previewed as text; you may need to ZIP the HTML, have the recipient extract it, and open it explicitly in Chrome. Your transfer method's file-size limits still apply. No time savings are promised.

Audio is inspectable. Neutral labels reduce ordinary expectation bias and provide no encryption or secrecy. Response checksums detect ordinary edits/corruption but can be recomputed by a technically capable recipient. Counts describe returned files, not verified people or scientific significance. No automatic alignment, mastering, format conversion, bit-perfect playback or certification is claimed. The inspectable creator editions have no account, activation or DRM; included open-source license notices apply.

After trying the free edition, what would make the $19 version worth buying for you?