This is not a complete code log. It is a version record organized for general users: what capabilities were added, which boundaries were corrected, and why those changes were made.
All 32 public portal pages now have complete static translations, with language switching, browser-language detection, canonical URLs, and multilingual search markup. Switching languages keeps users on the current article or app page and does not rely on runtime replacement of page text in the browser.
The Voice Knowledge center expanded from six foundational articles to fourteen. New articles come from real product development and fixes, covering Android M4A compatibility, special-sound routing, real-time pitch confidence, cents training, local vs server-side privacy, web music scheduling, real-time singing fallbacks, and reproducible test protocols.
The homepage no longer leads with an app directory. It now prioritizes original articles on fundamental frequency, recording variables, and result stability, along with real development notes on mobile formats, unusual-sound routing, and local processing. It also adds clear paths for understanding how conclusions are formed, who the author is, and all nine apps.
Star Sea Stage became the eighth official app and is featured at the top of the Family homepage. A full introduction page was added on the main domain covering multiplayer karaoke rooms, video loading, real-time singing transmission, room cleanup, and content-rights boundaries.
The main domain gained a dedicated introduction for every official app, explaining how it works, data handling, and capability boundaries. Six original Voice Knowledge articles were also published, covering fundamental frequency, resonance, recording variables, result variation, browser audio, and responsible interpretation.
Reborn Through Sound became the seventh official app and was added to primary recommendation positions in the Family portal and other official apps. Now any analyzable sound can enter the fantasy world and generate reincarnation coordinates for that particular sound.
Added Family overview, Voice Guide, report-reading, privacy, FAQ, author, and update pages so the products have more than operational entry points and can fully explain methods, boundaries, and development context.
Inner Voice Tarot became the sixth official app. Family-entry ordering, official names, logos, and new-product notices were standardized across apps.
Added six tiers of rarity titles based on anonymous sample proportions. Primary timbre and three-timbre combinations are calculated separately; saved images show the rarer one at the top, and the report's proportion section now includes a full achievement card.
CVoice, EchoSea, Insight, and FrostNote moved to independent product domains, with pages and APIs using the same origin to reduce cross-origin request failures on Android and embedded browsers.
The displayed tendency now follows the highest final proportion among feminine, neutral, and masculine voice impressions. When the lead is very small, primary-timbre candidates fall back to the general pool so tiny differences are not amplified into exclusive labels.
Added non-speech sound routing to keep whistles and similar audio out of standard gender/age impressions and rankings; expanded Android M4A and AMR file selection and server-side parsing.
Added more complete spectral, prosodic, and resonance-related features to improve voice-tendency and timbre matching. The new dimensions participate in internal calculations without dumping complex raw data onto entertainment result pages.