Sound Swap 0.1.0 candidate

Open sound-swap-offline.html in a current browser. Try the built-in synthetic
example, or choose a recording (up to 120 seconds / 40 MiB) and your own sound
(up to 5 seconds). Save WAV exports mono 16-bit PCM. The single-file candidate
is intended for direct opening; that mode has not been browser tested because
the testing tool blocks file:// URLs. Source files are under source/; the
separate module version uses a normal static server and was browser tested.

Audio stays in memory on your device. No source waveform is mixed into the
replacement. Detection uses loudness, not word recognition: quiet pauses and a
clear voice work best. For a natural bark or meow, use a recording you have
rights to. The default is a procedurally generated digital meow, not an animal
recording. Long bursts repeat the clip; short ones compress it. Preview the
result to judge suitability.

This is recorded-clip conversion, not a live microphone or virtual audio device.
It is not an anonymisation guarantee: rhythm and pauses remain characteristics.

MIT applies to our code and synthetic examples. Input audio retains its own
rights. No third-party audio, customer data, accounts, or analytics are bundled.

Author checks: eight DSP tests, independent Python WAV decoding, and browser
metadata/decode/example playback/reset/settings flow. Native file selection
and browser download were not exercised because of a separate tool approval
incident. Independent human testing and customer acceptance are not recorded.

The base package and browser use are free. Tuning one agreed recording and
replacement clip, a reviewed output WAV and settings notes with one correction
is a separate US$49 proposal. Agree materials and delivery before paid work.
Contact: ignat0111@gmail.com. Exclusive rights are not promised.
