MiniMax Music3 Studio 3.8.3

Every engine runs on any x86-64 processor, with or without AVX2. The vocal separator (BSRoformer.cpp), recognition for karaoke (transcribe.cpp) and the audio.cpp server now ship a processor build for every instruction level and load the one this processor runs best, instead of the AVX2 build alone; earlier versions could crash on a processor without AVX2. The new builds are downloaded once after the update.

The MiniMax Music3 engine (CUDA 13, CUDA 12.9, Vulkan, the processor), the CUDA runtime and the other paths are unchanged. The Linux .deb and AppImage are built on Ubuntu 22.04.

MiniMax Music3 Studio 3.8.2

YuE2 Studio, ACE-Step Studio and MiniMax Music3 Studio are on one version again: 3.8.2 follows 3.8.0 here.

Fixed

• The engine starts when its port is taken or reserved: the engine listens on port 8086, and when another program held it - or Windows had reserved it for Hyper-V, WSL or Docker - every song was lost with "cannot bind 127.0.0.1:8086". The studio now checks the port before the engine starts and, when it cannot have it, gives the engine a free one; the engine log says which program held 8086 (its name and PID) or which reserved range it lies in, and a song lost to a port taken later says so in the window's language.
• An update that cannot replace a file in use leaves the earlier one in place: when the new file could not take the place of one a running program holds, the earlier file stayed moved aside and the program lost it; it is now put back, and a file moved aside in this run is not swept until the next start.

MiniMax Music3 Studio 3.8.0

Fixed

• A runtime of 3.7.0 is replaced by today's pin: llama.cpp, cuBLAS and the other runtimes 3.7.0 unpacked keep the file names of today's builds, and were taken for them; each download now leaves a record beside it (its address, hash and files), and a runtime from before records counts as today's only when the file that tells its build apart has the pinned size. An earlier build is fetched again, removing only the files its record lists and only once the new build has arrived. An install is marked as under way until its record is written, so a half-unpacked runtime is not taken for installed, also after the studio was closed in between.
• An update does not stop halfway on a file in use: a runtime that karaoke, the assistant or a separator still had open could not be removed or replaced on Windows, so an update stopped with "Access is denied" and left neither the old build nor the new one; such a file is now moved aside, the new one takes its place, and the old one is removed at the next start.
• An M4A of Apple Lossless goes into the library as FLAC: every M4A was kept as it was, and the window's player does not play ALAC; one plays as it is only when its audio is AAC. A WAV whose fmt chunk does not come first (after a JUNK, bext or LIST chunk) is read for what it is.
• Karaoke keeps the first word after a filler: a word that starts where a segment the model filled with "[Music]" ends was left out with it.
• Changing the cover look no longer stalls the window: switching between Pattern, Photo for the style and AI generation writes the new cover into every track, and the window read the whole library again after each one, with the replay data of every song: 5.2 MB for a library of 37 songs. The library list now leaves the replay data out (a re-render reads it from the song), 0.3 MB for the same library, and a burst of changes is read at most every two seconds (from YuE2 Studio #150).
• Models in a folder with a name outside the system's code page are found: on Windows the engine read its models folder through the system's ANSI code page, so a path with, say, "é" on a Russian system or Cyrillic on an English one became "?" and the engine started with no models; it now runs with the UTF-8 code page and opens any path.
• ADPCM WAV files open: a WAV in ADPCM (a mastering reference, a training song) failed with "no decoder for this audio"; it is read now.
• The assistant starts on Linux with a glibc older than 2.38: the CUDA builds of llama.cpp need glibc 2.38, so on Ubuntu 22.04 and the like the assistant did not start; there it now runs the card through the Vulkan build, and its settings say why, as in Speech Studio.
• Programs started from the Linux AppImage no longer load its libraries: the file manager, nvidia-smi, the downloaded engine, llama-server, Whisper, the MIDI tool and the trainer were started with the image's library path and could stop on a newer system («undefined symbol»); they start without it, as in Dub Studio.
• A program downloaded without its execute bit runs on Linux and macOS: an archive packed on Windows carries none, and the program answered «Permission denied» or was left out of the unpacking; the studio makes it executable.
• Karaoke, stems and the measurements load on a computer with an old Visual C++ runtime: ONNX Runtime imports msvcp140_1.dll, which the studio did not check for, so a computer whose Visual C++ runtime predates it passed the check and ONNX Runtime then did not load. The Visual C++ runtime the engine and ONNX Runtime import now ships beside the engine, as in the other studios, and ONNX Runtime finds it there, also before the engine has started; the studio no longer downloads and runs Microsoft's redistributable installer, which needed administrator rights and the network.
• "Create" no longer seems to do nothing: when the song could not start - the writing assistant was not reachable, a field was missing - the reason was written at the foot of the form, below what the window shows, and the click looked lost. The form now scrolls to the reason.
• Mel-Band RoFormer stems on a Mac, and on the processor under Linux: the Mac had no RoFormer build at all; it now downloads the author's Apple silicon build of BSRoformer.cpp, which runs on Metal and on the processor where Metal cannot. The processor builds for Linux and macOS ship their libraries by their full version alone (libggml.so.0.15.1), so the program did not find the name it asks for (libggml.so.0) and did not start; the download now gives each such library that name beside it.
• Audio codes in the Create form are used: the service dropped the form's audio codes field, so the song was made from scratch, the language model ran its whole length, and the codes changed nothing; they now reach the engine, which decodes them and skips the language model, as a re-render does, and a limited LM KV cache is counted by the codes' length, not the song's.
• A caption with only some of its headings keeps them once: a caption with only "Global Metadata" (or two of the three parts) was taken whole into the first field, heading included, and the form then wrote "Global Metadata" twice; each part now goes into its own field, and a heading word inside a sentence stays text.
• Every window language is named in its own language: Settings - Appearance - Language wrote Russian in the window's language («俄语» in a Chinese window, "Russian" in an English one) and, in English, Chinese, Japanese and Korean too, so a reader looking for their language did not always find it; the list now reads Русский, English, 简体中文, 日本語, 한국어 in every window.
• Every audio file the studio reads is taken everywhere: the library import took only MP3, WAV, FLAC and Ogg, the mastering reference and training pickers each had a list of their own, and none had the formats the decoder reads since this version. The window, the uploads, the import and the agent's tools now take one list - MP3, MP2, MP1, WAV, FLAC, Ogg, M4A/MP4/AAC, AIFF/AIFC, CAF and Matroska audio - and a file the player cannot play as it is (CAF, MKA, AIFF, MP2, ADPCM WAV) goes into the library as FLAC, losslessly; the import hint names them. Opus is not read yet, and is refused with that list.
• A CUDA failure names the card it happened on: a song lost to a CUDA error now says the card, its compute capability, the driver and the engine's CUDA build, which tells a card the build has no code for from a driver too old for it; an engine start whose CUDA device is missing quotes what CUDA answered.
• A lost song names how the engine ended: an engine that crashed during a song was only "stopped"; its exit code is now read before the restart replaces it, noted in the engine log and told in words in the studio's language - an access violation, an instruction this processor does not have, a library that did not load, memory that ran out.
• The chosen compute device reaches the engine: since the upstream update the engine takes its device as --device and no longer reads GGML_BACKEND, so a card, Vulkan or the processor chosen in Settings - Local engine was ignored; it is passed as --device, and a device the engine does not have is named at the start with the ones it has. A failed song's reason is looked for in the whole engine log, not the window's last 512 lines.
• The statistics count what people ask for, not the window's own polls: the setup status, the computer's load, the activity, the running songs, the karaoke and assistant states and the engine start and log were counted as features each time the window asked for them by itself; they now carry x-hub-skip, as the shared standard says.
• The window keeps up with a slow service: the setup status, the engine start, the tools, LoRAs, training, the agent panel and the engine log were asked again on a timer whether or not the last answer had come, so a service answering slowly had the same request queued several times over and the window showed what was on disk minutes late; each is now asked again a pause after its answer, one at a time.
• A download from a server this network cannot reach fails in a minute and a half, not six, and names the server: a connection that never opened - blocked, refused, timed out, no DNS answer - was tried eight times, 30 seconds each plus waits, before the setup said "could not start ... after 8 attempts"; it now gives up after three such attempts and says "could not connect to ", so it is clear whether Hugging Face, GitHub or NVIDIA is out of reach.
• "Use models I already have" says what it took and what it left, and why (#32): picking a folder of downloaded models only refreshed the list, and a file it did not take was left without a word. It now says how many files it took and names each model file it left: one with the studio's name but another size (another build - the Light sets' language models on Hugging Face are now imatrix builds of other sizes, the studio takes the pinned ones), a GGUF of none of the studio's sets, or one it could not link or copy.
• A song finished while a model or LoRA arrives is kept: the engine is restarted to find a new model or LoRA, and it was restarted as soon as it stopped rendering, before the studio took the finished song, which was then lost ("mm-server completed the job, but the studio could not safely import its result"), and the songs still queued on the engine with it. The restart now waits until the studio has taken every song it sent.
• A graphics card that is not there is said, not left to the processor: a card chosen in Settings - Local engine that the computer no longer has made the engine go to the processor without a word; the start now stops with the reason on screen, the card list shows the missing card so another can be chosen, and an engine running on the...

MiniMax Music3 Studio 3.7.0

Added

• Cleaner vocals with Mel-Band RoFormer. Settings - Models - Stems now offers Mel-Band RoFormer voc_fv6 (Q8_0), Dub Studio's separator run by BSRoformer.cpp, beside HT-Demucs: it splits a track into the vocals and the instrumental, more cleanly, and the vocals of training are separated with it too; on an NVIDIA card it runs on the engine's cuBLAS, elsewhere on the processor. HT-Demucs stays the default for six stems (YuE2 Studio #108).
• LoRA from your own folder, without a copy. LoRA page - Add from folder: every LoRA of a folder you keep for all your programs (ComfyUI, a trainer) and its subfolders is taken in place - a folder with adapter_config.json is one LoRA, elsewhere each .safetensors is one - by a hard link, copied only from another drive, and the studio says which; agents use lora_import_folder (YuE2 Studio #98).
• A saved stem names its song's tempo and key. A vocal or any other stem saved for a DAW is named like Song - vocals - 109 BPM, D major.wav, as the song records them (YuE2 Studio #108).
• A unique pattern for every song. Settings - Covers - Pattern - Unique pattern of each song: the shapes come from the song's lyrics (a song without lyrics, its title) and the colours from its style, out of the pattern style's own palette, so songs of one style share their colours and new lyrics draw new shapes; a changed style or text draws the pattern again. The choices now read Pattern, Photo, Generation; a photo for the style stays the default (ACE-Step Studio #52).
• Your own agent CLI writes the words. Settings - Models - Writing assistant - Agent (CLI): Claude Code, Codex, opencode, Antigravity or Muse Code installed on this computer writes on your own account with it, one hidden run per draft, without tools; the panel shows which are installed, how to install and sign in, and tests one with a short draft. Antigravity runs only sandboxed. Agents see the CLIs with assistant_cli_agents (from YuE2 Studio #112, thanks to GuilhermeAmado).
• An agent reaches everything the studio does. A test now holds every route of the studio to a tool or to a written reason, and the agent gains what it lacked: the network proxy (get, set, test), access from the network, the Activity journal, the look of tracks without a cover, photo and pattern covers and photo search, video background search, a song's note, its own MIDI file, a playlist by itself, the graphics cards and the downloadable parts' state.
• Lyrics in a video clip, found where they are. The video screen's Text tab always shows the lyrics overlay: a song without karaoke timing gets a button that makes it there, and the overlay turns on when it is done; without karaoke set up it says where to set it up. Until now the overlay section was hidden for such a song (#118).
• Fade in and fade out. The create page's output settings take seconds for a fade at the start and at the end of the made song, so an ending no longer stops abruptly; the studio applies them to the song it keeps, Reuse Prompt and a re-render keep them, and agents give fade_in and fade_out.

Fixed

• Karaoke on an instrumental no longer hangs on "downloading": the studio checked the song after fetching the recogniser and left that download's line running when it then refused the track; the song is checked first, and a finished download ends its line.
• A long step no longer gets the engine restarted (YuE2 Studio #129): an engine inside a long step does not answer its health check in time, and the studio took that for a crash and restarted it, losing the song; while a job is running and the engine process is alive, it is at work.
• Auto stays on the graphics card after a silent engine exit (YuE2 Studio #127, #128 by ramishi): an exit with no device error in its log (out of memory, a kill, a slow start) no longer moves Auto to the processor for the rest of the session.
• The create form keeps its words (ACE-Step Studio #54): the name, the style and the lyrics stay across restarts, as the length and the file format do, until Reset clears them.
• The training page says where to install the card runtime. When preparation measures tempo on the processor, its notice now names the way to the graphics-card ONNX Runtime for this machine's card: Settings → Models → Optional extras → Split into stems → Graphics card on NVIDIA, Karaoke timings → Graphics card on AMD and Intel (DirectML).
• A computer's memory as its box says it. The studio counted RAM in 10⁹-byte gigabytes, so a 32 GB computer read 34 GB (in the issue form too); memory is now counted in 2³⁰-byte gigabytes, as video memory already was, and so are the model sets' needs.

Changed

• Karaoke's Parakeet on transcribe.cpp. Parakeet v3 and Parakeet Ultra run as GGUF Q8_0 on transcribe.cpp, the engine Dub Studio and Speech Studio use, instead of ONNX Runtime: its CUDA build on an NVIDIA card with a CUDA 13 driver, its Vulkan build on any other card (and on Linux without the system's CUDA 13 cuBLAS), its processor build without a card; on a Mac Metal on Apple silicon and the processor on Intel. The files have Dub Studio's names and folders, so one program's copy serves the others, and the model leaves the graphics card when a recognition ends, so the card is the music engine's again. A saved fp32 choice becomes Parakeet v3 (full weights are not kept), and the ONNX Parakeet folders of 3.6 and earlier are removed at start.
• Whisper on the graphics card keeps int8 weights (int8_float16) instead of float16.
• Audio to MIDI and the training pack are one download. The MIDI transcriber, the trainer and the captioner come from one HOT-Step build that shares its graphics-card libraries, in one folder: 133 MB once instead of 126 and 132 MB apart, and installing either one installs both (YuE2 Studio #104).

Also in this release (prepared as 3.6.1, not published on its own)

Fixed

• An agent sees the window as you do. Through the studio's MCP it now sees folded sections (open or closed) and the options of open lists, reads a field by its own label rather than by the options inside it, and a field it fills while the window is in the background saves as after your own click.
• LoRA preparation says which song does not fit and why. A song longer than the model's 24576 tokens (prompt, score and audio together) passed preparation and failed at its end with "prepared manifest failed reader roundtrip"; the trainer now names the song and its sizes, and a roundtrip failure carries the reader's reason (#110).
• Karaoke and stems on a Mac. The Mac build downloaded the Windows ONNX Runtime and could not load it; it now takes Microsoft's macOS arm64 build, runs Parakeet and the separator on the processor, and the app may load that library (it is signed by Microsoft, not with the app). Whisper and the CUDA and DirectML parts are not offered on a Mac, where they cannot run (YuE2-Studio #105).
• A download Hugging Face does not answer says why and what to do. The error kept only "error sending request" and lost the cause (the name not found, a timeout, a refused connection); it now names it, and when the model comes from Hugging Face it points to Settings - Network, which can take models from the HF-Mirror instead.
• Karaoke that cannot start says why and ends. An ONNX Runtime the studio could not load stopped the recogniser with a panic: the karaoke stayed "running" for ever and every later try answered "Mutex poisoned" until a restart. The library is now loaded before any work and a failure is that work's error, with the library's path; a recogniser that stops ends its karaoke as failed (YuE2-Studio #106).
• The song details button is back on wide windows. 3.6.0 hid it on the create page whenever the details could sit beside the list; it shows at every width again (YuE2-Studio #99).
• The Linux AppImage starts its engine again. The AppImage put the MIDI tool's older ggml library on the path every program inherits, so the engine loaded it instead of its own and stopped at once ("undefined symbol: ggml_prec_set_acc"). Every program the studio starts now loads the libraries beside it first: the engine, the writing assistant, the trainer, Audio to MIDI and the karaoke recogniser (YuE2-Studio #97). Thanks to angelside, who found the cause.
• The writing assistant works on Linux. It downloaded the Windows builds of llama.cpp and could never start; it now takes llama.cpp's Linux builds for the processor, Vulkan and CUDA (YuE2-Studio #114).

MiniMax Music3 Studio 3.6.0

Update from 3.5.5.

Added

• A library as roomy as a music service's. Compact rows show three times as many songs; the tabs count their songs and the filters say how many of how many are shown. Filters pick where songs came from (made in the studio, imported, stems and processing) and those with karaoke or processed versions; Recently played joins the orders. A right click on a song opens its menu, which gains Play next and Add to queue. In a selection the bar stays on screen while the list scrolls, Shift+click takes a range, Ctrl+A every song and Esc leaves it, and the chosen songs can be liked, queued or downloaded at once. Liked and Playlists say what to do when they are empty.
• Playlists to keep and play. A playlist's name and description change in place, its songs move by dragging or Alt+Up/Down, and songs are found and added on its own page; a song taken out comes back with Undo. Its picture is the covers of its songs, the same on its card and its page; the song playing is marked; Shuffle plays it in a mixed order, and a card has Play on its cover and a menu on a right click. Adding songs says which playlist they went to and which were there already. Back from a playlist returns to the Playlists tab. Agents queue songs with player_queue_add; playlist_update takes the whole order.
• Winamp plays the studio's playlists. The studio's shuffle and repeat go into Winamp and come back; LOAD LIST and ADD open the studio's playlists, the liked songs and the whole library, and SAVE LIST writes over the playlist the music came from or into a new one.
• A track's file carries how it was made. Besides the title, lyrics and cover, the file now holds the song's note as its comment, its whole style, and its generation settings (a JSON document of seed, models and options, without paths on your computer) in fields of the studios' own. Importing such a file back fills Reuse Prompt with what it was made with, in the studio that made it; any file brings its title and lyrics, and the library says how many files came in. Tracks made before get theirs at the first start. Settings › Privacy turn it off, and every stored track is tagged again (ACE-Step-Studio #50).
• Several tracks to MIDI at once. Choose songs in the library or on the create page and press To MIDI: they are turned into MIDI one after another, Studio tools shows which track it is on and how many are done, Stop ends the set, and a track that fails is named while the rest go on. Agents give song_ids to midi_transcribe (YuE2-Studio #94).
• Sorting out a long list of songs. A song's row has a Delete button beside Add to playlist, and the chosen songs of a selection go to a playlist in one go. The Library's All and Liked tabs have the create page's search, filters, selection, Like, Add to playlist and Delete, and a details button that opens the song's details over the page and closes them again (ACE-Step-Studio #48).
• OGG and FLAC songs can be imported into the library, beside MP3 and WAV, and turned into MIDI, stems or karaoke like any other (YuE2-Studio #94).
• The version in the window title and under the studio's name, with "Portable" on a portable copy, so a copy waiting to be updated tells which it is (ACE-Step-Studio #40).
• Every settings slider takes a typed number. The value beside a slider is a field: type 0.40 for a LoRA's strength, 2:10 for a duration or 6 for an equalizer band, and Enter sets it, kept within the slider's range (a comma works as the decimal point, Escape puts the value back). On LoRA strength, the creation settings, audio processing, the equalizer, the Winamp size and the video clip editor.
• Any engine launch flag and a hand-built engine. Settings - Local engine take one line of extra flags for the engine, split as a shell would and placed after the studio's own, and an engine executable of your own build, picked in a file dialog and checked when saved. The restart button saves the page first and starts what it shows; an old engine still holding the port is waited for and refused instead of being reused with its old flags. Ported from @wuyue92tree's change to ACE-Step Studio (ACE-Step-Studio #46).

Fixed

• "Instrumental" is not a genre. A style that starts with "instrumental" no longer files the track under the genre Instrumental in its file; the next phrase that names music is taken, or none.
• An installation under Program Files downloads the engine's libraries. The studio put NVIDIA's cuBLAS beside the engine, where Windows lets no program write without administrator rights, so an installed studio stood at "Downloading the engine library - 0%" for ever. Where the engine's folder cannot be written, the libraries now go to the studio's data folder and the engine finds them there (ACE-Step-Studio #49).
• Stems, karaoke and Audio to MIDI work on Linux. The studio downloaded the Windows builds of their runtimes on Linux and stopped there without a word. It now takes Microsoft's and NVIDIA's Linux builds of the same versions - ONNX Runtime with CUDA, cuBLAS, cuDNN, cuFFT, cuRAND - and Purfview's Linux Whisper, runs the separator and recogniser on an NVIDIA card there, and the Linux packages carry the MIDI transcriber (YuE2-Studio #95).
• A download that cannot start says so in its panel instead of leaving the bar where it was.
• "Use models I already have" takes only the models in the folder. A file was taken for every model of the same size, and many models of one architecture weigh the same to the byte, so four files came in as a dozen models with the wrong weights under their names; a file under another name is now taken only when its contents are that model's (ACE-Step-Studio #40).
• One version everywhere. The studio's version is written in desktop/src-tauri/tauri.conf.json alone: the window, the service, the interface, the project page and the release all read it there, and the studio names itself to the sites it asks with it instead of 0.1.0.
• An agent's command to a closed window fails at once. A window that closed without saying so kept the agent waiting the full five minutes for its answer; the command now fails as the window goes, and the next call reaches another open window or says none is open.
• The window gets the microphone it asks for. WebView2 left the request with no answer and no prompt, so recording in the audio editor waited for ever; the studio's own page is now allowed the microphone, nothing else is.
• A song lost to the graphics card names the error. "The graphics card failed" now carries the engine's own CUDA or Vulkan error, and a song lost to an engine that ended without a word says the last thing it wrote, so the cause can be told apart instead of guessed.
• A recording that will not play says why. The player of a training song or a stem turned to Pause and stayed silent when the window could not play the file; it now stays on Play and names the reason the window gave (YuE2-Studio #75).
• Training songs play in the window. The player of a song in a training set asked the window's own page for the recording instead of the studio, so in the window it stayed silent while a browser played it (YuE2-Studio #75).
• Access from the network can be turned on from a browser on the studio's own computer, not only in its window, so a studio on a server without a desktop is reachable; and access turned on by hand in studio-settings.json without a key now gets one when the studio starts instead of staying on this computer only (ACE-Step-Studio #47).
• Downloads that stall come back. A connection that stopped sending held the download at the same byte for ever, with no error and no retry - most visibly the engine libraries from NVIDIA, which kept the start screen at "Downloading the engine libraries - 0%" (YuE2-Studio #92). A silent connection is now dropped after a minute and the piece is fetched again, and a server that refuses a file says so at once instead of after two minutes of retries.
• Only the studio's own window and agents can use it. A web page open in your browser could send requests to the studio on this computer, and the studio answered any site; it now refuses pages of other sites, and requests that reach it under another site's name (DNS rebinding), as Dub Studio does.

MiniMax Music3 Studio 3.5.5

Update from 3.5.4.

Fixed

• The engine starts on macOS. The bundled engine looked for its libraries in the build machine's folder and never started on a Mac; it now finds them beside itself (YuE2-Studio #79, YuE2-Studio #81).
• The writing assistant works with OpenRouter. Saving the model put the assistant back on a local provider, so with a key and a model chosen it still said "not installed" and the wand opened the setup page; the choice now stays, and the providers page no longer undoes a choice made on the models page (YuE2-Studio #80).
• Imported tracks show their real length. An MP3 was measured as if it were 128 kbps, so a 320 kbps file showed two and a half times longer in the library than in the player; files are measured from their own headers, and tracks already imported are measured again on start (YuE2-Studio #77).
• The song details button stays on wide windows. It disappeared above a certain width, and once the side panel was closed nothing opened it again (YuE2-Studio #84).
• The audio editor keeps the sample rate. It decoded every file at 44.1 kHz, so a 48 kHz song was resampled on opening and saved at 44.1 kHz; it now opens and saves at the file's own rate (YuE2-Studio #83).
• "Without stopping" stops when a song fails. It sent the same request again and again although each one failed - a deleted source song, a card out of memory - hundreds of times in a row; a failed song now switches it off and says why.
• A failed song names the cause. The message showed the engine's last complaint, such as "encode failed", instead of the first one that caused it.
• Models not downloaded yet are called that. "Not installed" next to a model read as a broken installation.

MiniMax Music3 Studio 3.5.4

Update from 3.5.3.

Fixed

• Exports from the MIDI editor are saved. Export WAV, Export MP3 and the MIDI download filled their progress bar and then did nothing in the desktop window; they now open the Save dialog, named after the song (YuE2-Studio #71, #72).
• Playback on Linux. The AppImage now carries the GStreamer libraries and plugins the window plays sound with, so playing a song no longer crashes on systems such as Arch (YuE2-Studio #70).
• Songs stay on this computer's engine. Settings from an older build could switch song-making to OpenRouter with no model chosen, and every song then stopped with "OpenRouter music requires a selected model"; songs go back to the studio's own engine, and OpenRouter can be picked for songs only once its catalog offers a model (ACE-Step-Studio #37).

MiniMax Music3 Studio 3.5.3

Update from 3.5.2.

Fixed

• Works offline. With the computer offline the window held every read and change until a network came back:
  a song stood at its last stage although it was made, and a song deleted came back when Wi-Fi was turned on.
  The window talks to the studio on this computer and no longer waits for a network.
• Describe by ear hears FLAC songs. Since songs are kept as FLAC the captioner refused them; they are handed
  to it as WAV.
• WAV (32-bit float) is offered again beside FLAC and MP3, in the form and in Re-render.
• Audio to MIDI keeps the size you pick. The models page put the choice back to medium while a download ran,
  so small and large looked impossible to get; the largest size downloaded is now the one offered first, on the
  models page and on the tools page.
• Send to the form keeps the song's seeds. The sound seed and the planner seed came to the form empty, so the
  same song could not be made again; both come along now and the random switch is turned off.
• Linux packages run on Ubuntu 22.04 and other systems with glibc 2.35 or newer; they needed 2.38.

MiniMax Music3 Studio 3.5.2

Update from 3.5.1.

Added

• News strips. Short notes from the author appear as a strip across the top of the window, in the
  window's language, with a link and a close button; the first one is about Dub Studio, the video dubbing
  studio. A closed strip does not come back.

Fixed

• Quitting and updating ask in the studio's own language. The question about quitting while a song is
  being made and the update offer are now part of the window, in its language and look, instead of English
  system dialogs. Closing the window a second time still quits if the window stopped answering.
• The day's statistics leave when the studio closes instead of at its next start.
• A model set put together by hand is counted as it was when the song started, not as it was when the
  song ended.
• Statistics count finished songs by job as well as by song, so the share of failed songs compares
  like with like.
• Linux builds are marked experimental in the README, with an invitation to Linux users to polish them
  and send pull requests.

MiniMax Music3 Studio 3.5.1

Update from 3.5.0.

Fixed

• Statistics name the model set in full. A set put together by hand is reported by its parts, and a day
  on several sets reports each with its number of songs; a failed song reports its reason, with paths, names,
  links and quoted text cut out on this computer before it leaves.
• An agent's Play no longer takes over the player while you listen: a song you are playing is neither
  switched nor paused, and the agent is told you are listening.
• A like is a heart everywhere: the song list and its Liked filter showed a thumbs-up while the
  player and the side panel showed a heart.
• On the start screen the statistics checkbox, its label and What is sent stand on one line in one
  colour; the link sat lower than the label and in another colour.
• The notes of 3.4.0 and 3.5.0 in News are laid out like the earlier ones: a summary line, then each
  change as a point with its name in bold.