discoDSP has released two free apps: WavePrompt, a local text-to-music generator for macOS and Windows, and Roboto, a singing robot voice synthesizer for macOS, Windows, Linux and iOS.

WavePrompt

WavePrompt is a free app that generates loops, vocal phrases and full songs from a text prompt. It runs the YuE2 music model on your own computer, on the Apple Silicon GPU on a Mac or an NVIDIA or Intel Arc GPU on Windows, with no server, no account and no upload.

Describe a sound and WavePrompt renders it locally. Loops are cut to exact bars at your tempo, vocal phrases sing your own words in nine languages, and songs run from intro to outro. A piano roll lets you draw or generate the melody and chords the result follows, and the latest update 1.0.6 adds Import Memo, which turns a hummed or sung recording such as an iPhone Voice Memo into a melody of up to 16 bars with matching tempo and key.

Features

• Bar-accurate loops from 1 to 16 bars at 60 to 180 BPM, in any major or minor key and in 4/4, 3/4, 6/8, 5/4 or 7/8.
• Sample-accurate loop starts with nudging, so loops drop straight into a DAW grid.
• Sung vocal phrases from your lyrics in English, Chinese, Japanese, Korean, Spanish, French, German, Italian or Portuguese.
• Eleven voice types including breathy, powerful, deep, raspy, falsetto, whisper, choir and rap, with optional vocal isolation via Demucs.
• Full songs from 15 seconds to 4 minutes, with vocals or instrumental, ending on a planned outro. Lyrics are checked with Whisper.
• Piano roll to draw a melody and pick a chord for each bar, or generate them in the chosen key with density, register and chord options.
• Import Memo turns a hummed or sung recording into a piano roll melody.
• 34 genres, 14 moods and 11 instrument sets, plus a style strength setting.
• Fixed seeds to repeat or vary a result, and a library that keeps every render with its recipe.
• Every result comes with a MIDI file of its score, with melody, chord, tempo, key and section marker tracks.
• Export to WAV (48 kHz, 24-bit), AAC or MIDI, with tempo matching by the Rubber Band Library.
• Works offline after the first model download.

System requirements

• Mac: Apple Silicon (M1 or later), macOS 26 Tahoe or later, 16 GB of unified memory (32 GB recommended for full songs).
• Windows: Windows 10 (version 2004) or Windows 11, 64-bit, NVIDIA GeForce RTX 30 series or newer with 12 GB of video memory or an Intel Arc GPU, 16 GB of RAM (32 GB recommended for full songs).
• About 10 GB of free disk space for the models and the engine.

Music is generated with YuE2 by Multimodal Art Projection. Model weights are licensed under CC BY-NC 4.0 with individual creator permission, and generated audio is yours to use under that permission.

More information: discoDSP / WavePrompt

Roboto

Roboto is a free singing voice synthesizer. Write a melody in the piano roll, type a word on each note, and Roboto sings it. It runs as a plugin or standalone app on macOS, Windows, Linux, iPhone and iPad.

The voice is synthesized with no samples, from a glottal pulse source, five formant resonators and shaped noise for consonants. Roboto sings in English, German, French, Spanish, Japanese, Russian, Finnish and Swedish, and the latest update 1.0.2 adds video export of the animated robot, waveform and lyrics with the song.

Features

• Piano roll where every note carries a word, with snap, zoom and a marker to sing from.
• Text tab that holds the whole song as text, so songs can be typed or pasted.
• Musicalize button that gives plain lyric lines a pentatonic melody, one note per syllable.
• Automation curves along the song for volume, pitch bend, vibrato depth, breath, formant shift and chorus.
• Words are turned into phonemes by built-in rules for each language. Phonemes can also be typed directly, for example S-T-R-IY-T.
• Ten voice presets: Default, Robot, Soprano, Bass, Child, Whisper, Choir, Opera, Giant and Alien.
• Transpose, tempo with host sync, vibrato, jitter, breath, consonant level and length, formant shift and bandwidth, reverb and double track.
• Animated robot face with LED eyes and a mouth that follow the voice, plus a live waveform and lyrics that light up in time.
• MIDI keys from C1 sing from the matching bar while held, and MIDI controllers drive volume, vibrato, breath, formant shift and chorus.
• Import MIDI, karaoke MIDI (.kar), UltraStar, LRC lyrics, SRT subtitles and song text, from the menu or by dropping the file on the window.
• Export to WAV, MIDI, karaoke MIDI, UltraStar, LRC and SRT, with lyric times that match the WAV.
• Video export (MP4, 1280x720 at 60 fps) on macOS, Windows and iOS.
• Built-in songs in every language.
• All voice parameters can be automated by the host.

Formats and system requirements

• macOS 12 or later (Apple Silicon and Intel): VST3, AU and standalone.
• Windows and Linux (64-bit): VST3 and standalone.
• iOS 15 or later (iPhone and iPad): AUv3 and standalone.

More information: discoDSP / Roboto

Read More

AudioPrism, the Windows app for offline stem separation, per-instrument MIDI export and piano practice, is $59.99 until October 31, 2026 (regular $79.99).

Version 2.0.1 exports a piano, guitar, bass, vocal or other stem to its own MIDI file, lined up with the song in a DAW. KeyPrism turns any song into a falling-notes piano lesson.

{PRODUCT-HEADER-38008-audioprism}.

Read More

AudioPrism has released version 2.0.1 of AudioPrism, its Windows app for offline stem separation, MIDI export and piano practice.

AudioPrism separates audio offline on your own PC with Demucs into an acapella, an instrumental, any single instrument or all six stems (vocals, drums, bass, guitar, piano, other). It is fastest on an NVIDIA GPU and falls back to the CPU without one.

It pulls lossless WAV from audio files or from YouTube, SoundCloud and direct links. KeyPrism, built in, turns any song into a falling-notes piano lesson with note names, fingering, slow-down, looping and hands-separate practice.

What's new in 2.0.1:

• Stem to MIDI: exporting a piano, guitar, bass, vocal or other stem now writes that instrument's own notes to its own MIDI file. Previously every stem came back as a piano cover of the whole song.
• The MIDI lines up with the song: drop the song and the .mid into a DAW at the same start and they match, at the song's tempo.
• Tighter piano covers: a new beat tracker fixes songs whose beat was misread, and covers now carry dynamics and sustain pedal, so they play like a pianist in any DAW piano.
• Hands split by what they play (tune right, bass left) instead of at middle C.
• "Full cover" replaces the Balanced and Real piano options, which produced the same part.
• KeyPrism no longer plays clicky guide notes under the original track.

{PRODUCT-HEADER-38008-audioprism}.

Read More

Euphona has released Euphona Studio, a browser-based audio production service covering mastering, stem separation, restoration, vocal tuning, mixing and binaural spatial audio.

A track is analysed once (loudness and dynamics over time, spectral balance, key and tempo, transients and damage), and each process works from that shared analysis rather than starting again from a bounced file. Results are played back level-matched against the original.

Features:

• Mastering with a -1.0 dBTP true-peak ceiling by default and a measured loudness report.
• Stem separation into up to six stems.
• Restoration for hum, clicks, noise, clipping and room.
• Vocal tuning (pitch correction).
• Mixing from the user's own stems.
• Binaural spatial audio for headphones, built from stems.
• A developer API running the same engine.

Pricing:

• Free plan gives 10 minutes of processing to hear results.
• Creator at $15 per month (120 minutes).
• Professional $29 per month (400 minutes).
• Max $79 per month (1,500 minutes).

Free browser tools (loudness checker, BPM and key finder, spectrum analyser, stereo width analyser, clipping detector and vocal remover) need no account.

Read More

Spongefork has released ShadowTag v2.3.1, an update to their free audio metadata editor for macOS.

Spot to Timeline, introduced in 2.2, now sends files from a ShadowTag playlist onto the timeline of eleven applications, adding REAPER, Cubase, Nuendo, Studio Pro, Adobe After Effects, and FMOD Studio, with clip colors carried across when possible.

The update also adds Relink Files, VST-XML copy for Cubase, Nuendo, and Studio Pro, options for Detect Loudness and Organize Keywords, and fixes that keep chapters, XMP, iXML, and non-Latin text intact when tags are saved.

Also in 2.3:

• Detect Loudness opens a dialog for scope, destination (loudness tags, BEXT, ReplayGain), skipping measured files, and album gain, and now measures Ogg Vorbis, Opus, Matroska, and MXF files, with surround measured by its actual channel layout.
• Converting 5.1 to stereo mixes down every channel, and FLAC and Ogg keep up to 8 channels.
• Markers can be copied and pasted between files.
• Show in Playlists lists every playlist that holds a file; Search Results and a new Folder column persist across relaunch.
• Tag saves keep MP3 chapters and XMP, MP4 media kind, multi-line comments, the iXML fields ShadowTag doesn't display, and accented and non-Latin text in BEXT, iXML, chapters, and markers.

Read More

sonicLAB has released SPATAI v3, a utility software that offers Large Language Model AI assistance to move sound objects in a virtual 3D acoustic space. It translates in real time the movement of multiple objects in Cartesian or polar coordinates and sends this movement data to other 3D audio spatialization software.

SPATAI v3 update adds MCP support:

• SPATAI now can be operated directly from Claude Desktop or ChatGPT.
• The user can select a preferred model and submit the prompts from the chat window and the LLM takes care of the communication with SPATAI through the dedicated MCP server.
• No Cloud LLM access is needed, no tokens are spent.
• Due to the chat window's persistent memory, the user can keep refining the prompt step by step. All requests are linked.
• Currently supported destinations are: Amadeus Acoustics, Timax2, IEM, IRCAM Panoramix, IRCAM Spat, L-acoustics, SpatRevolution, YamahaAFC, Holophonix.

This is a free update. Existing users should can replace the app files and check for the instructions on the manual on how to activate the MCP server on Claude Desktop or ChatGPT. New users can purchase for £20 pounds from the sonicPlanet webshop.

YouTube/YGpIjWT9Nis

YouTube/lbpW_srdlO4

Read More

Audible Genius has updated Syntorial, its course that teaches synthesizer programming by ear, to version 2.3.500. The free update for macOS, Windows and iPad changes how Syntorial's synth lesson packs fit into its core course, The Essentials.

Previously, a synth lesson pack replaced the course's On Your Own lessons. Its videos now sit inside them instead. When a user reaches an On Your Own lesson, a Translate to Your Synth panel shows which of their synths has a video for that concept, and tracks which videos they have watched.

The update also adds:

• A redesigned Lesson Packs window that shows what each pack covers before download.
• A View Lessons screen listing every lesson in a pack, with a short description of each.
• The option to buy lesson packs as paid add-ons to the full course.
• Optimized downloading and minor bug fixes.

Syntorial's existing synth packs (Serum, Massive, Sylenth1, Z3TA+ 2 and Minimoog Voyager) remain included at no extra cost. The first paid add-on, the Nord Stage 4 Lesson Pack taught by Marc Larochelle, was released alongside the update.

Pricing and availability (USD):

• Syntorial 2.3.500: free update for existing owners, from the user's online account or through the in-app update notice.
• Syntorial: $129.99, with a free demo.

Read More

IrisOrbit has released version 1.2.2 of PolyArp, a MIDI arpeggiator that turns a held chord into two independently controlled parts.

Pair a slow bass pattern with a faster upper line, combine straight and triplet rhythms, or send each part to a different synthesizer.

PolyArp works with software instruments and hardware synths, either standalone or inside a compatible DAW. Both parts can follow the whole chord, divide its lower and upper notes, or use an adjustable keyboard split.

Key features:

• Independent rhythms: set a different rate, direction, octave range, gate and velocity mode for each part, with separate MIDI output channels.
• Draw your own patterns: move and resize notes in a shared loop from 1 to 16 beats, using grid snapping or free timing. One part can play a drawn pattern while the other follows a regular note division.
• Shape the phrasing: Humanize, Expression and Articulation control dynamics and note lengths. Automate these controls and the gates to develop a phrase over time.
• Perform with MIDI controls: map knobs, pedals and aftertouch to musical parameters, including several parameters at once. Assign Release to a pedal or button to clear the current chord while keeping Latch enabled.

User presets recall the complete setup, including MIDI mappings and routing. PolyArp generates MIDI; the connected software or hardware instrument provides the sound.

Pricing & Availability

PolyArp 1.2.2 is available for macOS, Windows and Linux in Standalone and VST3 formats, with AU MIDI FX on macOS and LV2 on Linux.

A perpetual personal license costs EUR 15 in Europe or USD 15 elsewhere. A 7-day trial includes all features, starts after email verification inside the app and requires no payment card. An interactive rhythm demo is also available on the product page.

YouTube.com/watch?v=yF2GsyUz-8g

Read More

FluffyAudio has released FFT, a standalone spectral audio editor for macOS and Windows.

FFT combines spectral editing and audio restoration with machine learning tools for stem separation, speech enhancement, pitch and transcript editing and harmony generation, across 50 modules. All processing runs locally, and the optional machine learning models work offline once installed.

FFT is available at an introductory price of €99 / $99, 50% off the regular €199 / $199, until October 20. A free 15-day trial is available, no card required.

YouTube.com/watch?v=0JlvWeCD9AA

Read More

ProSound has updated AudioScout to version 1.1.

The standalone application helps sound editors, sound designers and podcast producers search and audition sound-effects libraries stored on local folders and external drives, collect sounds in a Queue and export them to a project folder.

Version 1.1 offers direct DAW workflows on supported macOS setups:

• Pro Tools Spot to a selected track and cursor position.
• Nuendo sound insertion and ordered Queue delivery through the bundled bridge.
• A dedicated REAPER bridge with one-time setup and automatic startup.

Pro Tools Spot requires Pro Tools 2025.6 or later; the macOS package includes the required interoperability runtime components.

Additional features include search by filename, folder context and supported embedded metadata, waveform scrubbing, region selection, Varispeed/Pitch auditioning, Reverse and Queue As Heard. Original sound files remain unchanged.

Pricing & Availability

AudioScout supports Windows 10/11 (64-bit) and macOS 13 or later on Apple Silicon. Local search, audition, Queue and folder export work without a supported DAW. The core local-library workflow works offline and requires no AudioScout account or subscription.

AudioScout is available for a one-time launch price of $10.

YouTube.com/watch?v=E1utRIZQ-Gc

Read More

Q.Audio has released Beat Warper Pro for Windows, a beat alignment tool built for DJ/Producers who need a fast and reliable, high-quality beat-warping to the DAW grid.

Beat Warper Pro is a tool for DJs, DJ-producers, and audio engineers who need instant, professional beat alignment. Whether you're working with AI-generated music, stems, or sample packs, this tool eliminates hours of manual warping in Ableton Live, Cubase, or any DAW.

Key Features:

Grid Alignment:

• Automatically detects all beats in your audio.
• Creates a mathematically precise grid.
• Aligns every beat to the grid with sub-millisecond accuracy.
• Zero time-stretching artifacts – pitch remains pristine.
• Export DAW-ready tracks (Ableton, Cubase, Logic, FL Studio).
• Automatically removes silence at the beginning of tracks.
• Preserves the attack of the first kick drum.
• Essential for DJs needing cued tracks.
• No more manual trimming in your DAW.

Universal File Support:

• Input: Any WAV, MP3, FLAC, M4A, AIFF, OGG.
• Output: 16 or 24-bit - 44.1kHz or 48kHz WAV (DAW-compatible).
• Handles any sample rate and bit depth automatically.

Batch Processing:

• Process entire folders of tracks in one click.
• Great for sample packs, stems, and entire projects.
• Progress tracking and detailed logs.

Professional Quality:

• Pitch preserved – no artifacts or pitch shift.
• Transparent processing – your audio sounds identical, just timed.
• 16/24-bit 44.1/48kHz output – ready for mastering.

Availability

Beat Warper Pro is available for Windows and is free / donationware.

Read More

Arvital Audio has updated AudioRoute to version 1.2.3. AudioRoute 1.2.3 is a maintenance release. Two of the fixes stopped recording from working at all for some users.

Changes:

• Windows: if the recordings folder or the Windows account name contained accented or non-Latin characters, recording failed the moment it was started, while the folder was present and writable. An account named José, or a folder named 录制, was enough to break every take. Any path now works.
• MacOS and Windows: recording could stop working permanently after a DAW or plug-in host crashed, and stayed broken until the background service was restarted, with nothing shown to explain it. The connections a crashed host leaves behind are now reclaimed automatically.
• MacOS and Windows: a recording that fails partway through, most often because the disk filled, is now reported instead of leaving a file that looks complete until it is played. Long takes are handled correctly as well: beyond roughly 3.1 hours at 48 kHz the WAV size field previously wrapped and the file became unreadable.
• MacOS: uninstalling from Settings now completes. In some cases it asked for the administrator password and then stopped the application that was waiting on the result, so nothing was removed and nothing was reported. Settings also stays reachable when the background service is not running, which is where the uninstaller lives.

AudioRoute captures the audio of chosen applications and puts it onto a track in any DAW as VST3, AU or AAX, into any application as a virtual input device, or straight to a WAV file. It runs on macOS 14.2 or later and Windows 10 or later. Existing users are offered the update automatically.

AudioRoute is 29 EUR until 15 October 2026 (Reg. 39 EUR).

Read More

discoDSP has released Discovery Pro 9.5, OPL 3.2, Corona 7.1, Bliss 3.26 and Retromulator 1.9.

Discovery Pro 9.5

Discovery Pro is a virtual analog synthesizer and sampler with 4 layers, 12 oscillators, 16 filter types including Nord Lead 2, Moog Ladder and Oberheim SEM models, SFZ/WAV/SoundFont/DWB/Bliss and Nord Sample Library support, PADsynth re-synthesis and Nord Lead 2 SysEx import/export.

Version 9.5 adds a zoom window for the Gate and Mod sequencers, with curve editing, a bipolar mode and an OSC1+2 pitch destination. The Nordic filter is rebuilt from the Nord Lead 2X hardware, and the filter menu is split into models and modes.

This update is free for all Discovery Pro 8.x and 9.x customers. Discovery Pro is available for Windows, macOS and Linux in VST, VST3, AU and standalone formats.

Bliss 3.26

Adds Gigasampler (.gig) import and an import picker for large disk images:

• Export Bank to Individual Programs writes every program as its own .zbp file.
• Velocity Curve in the Play Mode menu, saved per program.
• Third party imports run in the background with a progress bar and can be cancelled.
• Fixed Akai S3000 samples being read 42 bytes late, which added noise at the end.

Free for all Bliss 2.x and 3.x customers.

Corona 7.1

Improves oscillator hard sync and the oversampling decimation filter:

• Sync reset lands between samples, removing inharmonic sidebands around the slave partials.
• New half-band decimation filters, flat to 20 kHz with over 120 dB of alias rejection and lower CPU use.
• MacOS 12 or later is now required.

Free for all Corona 6.x and 7.x customers.

OPL 3.2

Adds a multi-operator mode with up to four modulators and four carriers per note on a pool of eight OPL3 chips:

• Eight preset algorithms plus Custom, edited on a Yamaha-style algorithm chart.
• Loads WOPL, WOPLX and OPLI banks and TX81Z, DX21, DX27 and DX100 sysex dumps, with over 9,000 new bank instruments.
• Master Volume with output meter, and a Saturator with Valve and Tape voicings.
• Sample-accurate MIDI timing, sustain pedal support and stuck note fixes.

Free for all OPL customers.

Retromulator 1.9

Adds an Oberheim Matrix-1000 and updates OpenWurli cores:

• The original firmware runs on an emulated Motorola 6809 board driving six CEM3396 voice models, oversampled 2x.
• Banks 2 to 9 load from the patch EPROM, and all 96 patch parameters have a MIDI CC.
• LFO 1 and LFO 2 can sync to the host tempo or incoming MIDI clock.
• The iOS firmware picker accepts several files at once.
• The Wurlitzer 200A OpenWurli core is updated to 0.9, with a field-solved pickup model, reed swing and air damping from per-note hammer tables, a new preamp solver, a tremolo with the real LED response, and the power amp and volume control networks from the schematic.

YouTube.com/watch?v=f8oNQlbjxbk

YouTube.com/watch?v=L1IRK9Ok4HA

Read More

OVNI Audio has released version 0.1 of Eyepiece and updated Telescope to v0.2.

Eyepiece

Eyepiece is a free macOS app that reads an Ableton Live, Logic, GarageBand or FL Studio project and writes a snapshot of it in text, to paste into the AI the user already works with. The AI gives the advice; Eyepiece only reads.

Eyepiece finds the most recently saved project on its own, or takes one dropped on its window. The snapshot lists what is in the session — the DAW and its version, tempo, time signature, tracks, clips, notes and devices — the shape of the arrangement, the mixer, the preset loaded in Arturia plugins, and findings, each with its number and the id of the rule that produced it. A coverage section lists what it did not read or could not verify, each with its reason. Detail stays out of the summary: the AI can ask for a code (eq:, notes:, plug:, ...) and Eyepiece copies only that block. The user sees the snapshot before copying it, and the button copies exactly what is shown.

Ableton Live projects are read in depth, including third-party plugins and the mix; the reader is verified up to Live 12.3.7. Logic, GarageBand and FL Studio projects are read at their first level. Eyepiece does not listen to the audio: every number in the snapshot comes from the project file. It needs no internet connection, sends nothing and has no AI inside. An example project comes with the app.

Features:

• Reads Ableton Live projects in depth: tracks, clips, notes, devices, third-party plugins and the mix.
• Reads Logic, GarageBand and FL Studio projects at their first level (FL Studio verified with 181 projects, from FL 10 to FL 2026).
• Findings with their numbers and rule ids, and a coverage report of what was not read.
• Detail on request, by code.
• Interface in English, Spanish and Japanese (the Japanese is a machine translation nobody has reviewed yet, and the app says so).
• No network access and no telemetry. The project is read, never modified.

Eyepiece 0.1.0 is free, for macOS 11 or later (Apple Silicon and Intel). It is closed-source.

Telescope 0.2

Telescope, the free and open-source audio analyser, has been updated to version 0.2.0.

The main fix: an open Telescope window no longer stalls the DAW. The update also adds a light theme, raises every axis label to 4.5:1 contrast or better, lets VERDICT reflow at 125% and 150%, and lets TONAL BALANCE compare against a chosen section of the reference track. What a file measures is unchanged: bit-identical to 0.1.0.

Telescope 0.2.0 is free and open-source (AGPLv3): VST3 and AU for macOS 11 or later, and VST3 for Windows 10 or later (x64).

Read More