- If microphone access is off, the app now tells you and takes you straight to Settings. - The screen stays on during live analysis. The microphone and playback stop when a call comes in, headphones are unplugged, or the app goes to the background. - Live analysis now receives unprocessed microphone audio for more accurate readings. - You can cancel while a file is being imported, and you are told why if an import fails. - Long files use less memory, and changing the language during analysis is safe. - Added a Manage subscription button on the License screen. Pending approvals, refunds and purchases made on another device are now reflected right away.

- Fixed the app freezing while opening a recording shared from another app. The status line now shows that the file is being imported.

- Buttons in the top bar and view row are now grouped by purpose. - Settings › About has been redesigned: app logo, version, update check and license in one place. - Information lines across the app are now shown as easy-to-read badges. - A light haptic tap now tells you when a long press opens a menu. - When there is no recording or the sound is too quiet, AI deep analysis now tells you right away. - Fixed a possible crash after sending an error report.

- Where very strong falsetto can't be told apart from mixed voice by sound, it's now shown as "Mixed or strong falsetto". - Whistle register is now detected on its own and shown in the range view. - The purchase screen scrolls when it is taller than the screen, so every button stays reachable. - Fixed text overflowing on small screens and with larger text sizes. - Fixed line wrapping in the help text.

- Tapping a feature that needs a license now opens a purchase screen that shows the subscription name, period, price, auto-renewal terms, Terms of Use and Privacy Policy first. A purchase only starts from the Buy button on that screen. - While you scroll, buttons no longer fire, lists no longer open and the keyboard no longer comes up. - Larger touch areas for How to use, Settings, the view buttons, the zoom buttons and the switches in Settings. The view buttons stay on the same row when you change views. - Live analysis: the blank band left between the graph and Start after stopping is gone, and the channel meters are easier to tap. - In Settings, typing the start of a model name shows the closest models right below. Model lists show the newest models first, and an empty list no longer opens an empty box. - Range view (loudness): the colour bar under the graph now matches the cells. - The register reading of falsetto recordings now holds steady even when EQ has been applied. - In 13 languages including Spanish and German, license and renewal texts that appeared in English are now translated.

The whole app has a new design. The measurements are gathered in one black row at the top (swipe it sideways), and the buttons above the graph switch between the six views. The four buttons at the bottom are Open file, See example, Live analysis and AI analysis. Touch and drag on the graph to see a red line and a tooltip with the note name, cents, frequency, accuracy and register at that point. The Range view can now show two pictures, Time held and Loudness. Time held shows how many seconds you stayed on each semitone as a bar. The Live analysis and Settings screens have been reorganised too. In Settings, tap an item in the list to open it. Fixed an App Store subscription license being turned off at launch or at midnight. Restore Purchases now checks once and shows the result once. With a license, no ad is loaded at launch at all. Also fixed an occasional crash report when quitting. Fixed the app showing Korean on a fresh install on a device set to another language; it now follows the device language. When the view buttons do not fit on one line, they wrap onto another line instead of being cut off, and the Formants view labels vowels with IPA symbols when the app is not in Korean. The Range view is back to the pitch × loudness grid. The button row at the bottom now reaches the edge of the screen, and the empty band at the top of Settings is gone. Sliders now change only after you tap to select them and then drag, so brushing past one while scrolling no longer changes it. Lists stop cleanly at the end without shaking.

A position bar with the current and total time now sits next to Play and Stop, and you can drag it to move around. The buttons are now in the order View, See example, Live analysis, Open file, AI deep analysis, How to use and Settings, and Zoom out, Zoom in and All sit at the top right of the graph. Tap the graph to see the note, pitch and register at that spot in a small bubble. Throat tension is now judged more strictly, so even a little extra tightening compared with your low notes is flagged. When you sing the same pitch several times, the app also checks whether you tighten more over time, using your first note as the reference. Each note appears as a bar in the Tension by pitch view, and the summary shows Tighter than at first.

Help text that was always on screen now sits behind the How to use button: the note under the graph, the live view guide and the history guide all open from there. Small text in the graphs, legend and summary chips is larger, and legend names are no longer cut off. The empty space under the graph when no file is open is gone, and summary chips scroll inside their row when there are many. The playhead now has a thin outline on both sides, so it stays visible over red areas. AI chat notices and errors show properly, and chat bubbles no longer overlap and fit their longest line. Settings have a cleaner layout, split into General, UI, AI, Memory, License and About, with help behind the How to use button at the bottom left. Korean text now wraps between words, and text that was cut off on narrow screens or left untranslated in other languages is fixed. The error window after an unexpected close now shows its full text. Dragging the graph or playing back redraws only what changed, so the app does less work. The microphone permission message now appears in English on English devices.

Tension by pitch now redraws for each file you open instead of keeping the previous result, and playback works again after you change the app language. If the server does not answer in time during AI deep analysis, the app now tells you what happened and what to try. The Save button in Settings has moved to the top right, and the long empty space below the settings is gone. When you pick a video from your photo library, the app now shows its progress while it comes in, and the picker appears in the app's language. Top-bar buttons and setting names are no longer cut off in English and other languages, and a long license status now wraps to the next line instead of being cut off.

If the app closes unexpectedly, the next time you open it you'll see what happened and can send a report right away. It follows the Report problems and Send error reports automatically settings. The back button on full-screen panels is now a plain text label, and a long account line on the License screen now wraps instead of being cut off.

You can now see whether you squeeze your throat on high notes from a single recording. There is a new view, Tension by pitch. It takes your lowest sustained chest voice in the recording as the reference and shows, bar by bar, how much more throat tension you use as you go higher. Until now tension could only be compared against your earlier recordings, because a relaxed throat sounds different for everyone. An easy low note from the same take, on the same day and the same mic, now serves as that reference. Just hold one easy low note and one high note for more than a second each. The harder the vocal folds are pressed together, the weaker the first harmonic becomes against the second. BEAN VOICE looks at that difference after removing the part shaped by the mouth and throat, so it can tell a higher note apart from a squeezed one. A bar past the dotted line means clearly more throat tension than your low notes. Under each bar you can see how much F1 rose; on the same vowel, that hints at whether your larynx rose with the pitch. Belting, where the folds are pressed together on purpose, also crosses the line, so listen as well before judging. Recordings with backing music are not judged. The summary row now has High-note tension, telling you at a glance whether your high notes are tighter than, similar to, or looser than your low notes. The AI analysis receives this result too. On small iPhone screens, a long summary row no longer squashes the graph.

You can now see whether your throat is tightening or tiring. Throat tension shows up in how hard and how abruptly the vocal folds close. When you tire or swell, the folds stop vibrating evenly, loudness gets uneven and the periodic structure blurs. The app reads those together. Because the resonance of the mouth and throat is removed first, the answer holds up even if you use EQ or a different microphone. What counts as comfortable differs from person to person, so the app never declares that you are tight. It compares the take against your own earlier recordings and tells you whether you tightened or loosened since then, matching the same pitch range across both takes. Voice range profiles can be overlaid. How loudly and how softly you could sing at each pitch is drawn for two days on one graph, so you see not only a wider range but where you could not sing out before. The point where chest turns into falsetto is compared too. If it moves up, you are carrying the mix further. Carrying power and the belt-versus-classical balance are shown alongside. Every record can carry a note of your own, because file names alone do not tell several takes of the same song apart. A search field filters the list as you type, by name, note or date. Comparing is no longer limited to two records. Pick several takes of the same song and the differences caused by the song itself disappear. With three or more, the ups and downs between them are drawn as a small line. On the live screen, press Stop and drag across the graph to read the pitch, loudness and register at any point. The compare and live screens used to appear in Korean in every language other than Korean and English. They are now fully translated. Fixed a crash when an iPhone was picked as a wireless microphone for live analysis. When the sound comes in from one device and goes out to another, the two are combined into one, and the app could die in the window where that combination is torn down as the iPhone connects. Live analysis does not send any sound out, so it now opens only the input side and that window is gone. The microphone picker is better too: an iPhone appears in the list as soon as you connect it, and re-picking the microphone you are already using no longer closes and reopens it. "Live" is now called "Live analysis".

Added measures taken straight from the vocal folds. By removing the resonance of the mouth and throat, a thin chest voice and a falsetto are now told apart clearly. Added the voice range profile: the pitches and loudness you can actually reach, on one chart, with the spots you stay in drawn darker. The app also measures how long it takes you to start a note, and whether your mouth shape follows the pitch on high notes. Pick two recordings from the list and tap Compare to see what improved between lessons, side by side. Tap Live to watch the note name and register of what you are singing right now. Nothing is recorded or sent anywhere. The app is lighter, and small layout glitches around the screen are fixed.

Fixed every place where a feature name was cut off on screen. - Names stay fully visible even at larger text sizes. - Button names are now shorter and clearer.

Your AI analysis conversation now lasts until you quit the app. - Leave the analysis screen and come back: everything you asked and heard is still there. - If you leave while waiting, the answer lands in the conversation and is there when you return.

Settings are now five tabs, and the app keeps a local record you can use as memory. - Settings: General, License, AI, Memory, About. - Pick the app language and a light or dark theme in General. - Name your coach and pick one of ten coaching tones. - Memory keeps measurements and conversations on this device only, with select-and-delete and delete-all. - Ask "am I better than last time?" and the coach compares against your earlier recordings. - The AI analysis screen is now a chat, and answers no longer stop mid-sentence.

You can now open a video straight from your photo library and read the voice in it.