M10 · Audio Production

transcription and reminders, loudness measurement to ITU-R BS.1770-4 / EBU Tech 3342 with true-peak, an offline DSP chain, non-destructive audio intelligence (stem classification, BPM, gain staging, cataloguing) and real procedural generation, stem separation (HPSS + centre), vocal tuning (pyin + pitch shift) and time-stretch (phase vocoder); neural back-ends are used automatically only if installed. Since the MASTER build of 2026-11-13 it includes the Universal Session Co-Pilot & Audio Pre-Processor 1.0.0 (7 of the 25 capabilities: copilot_analizza_sessione, copilot_prepara_sessione, midi_analizza, midi_elabora, copilot_stato, copilot_ambiente, copilot_libreria): automatic session detection with an Inbox watcher and a queue, track role recognition and standard names on copies, audio quality analysis, phase and alignment between microphones, acoustic fingerprints and duplicates, session quality control with a readiness score, a session workspace with a SHA-256 manifest, a MIDI engine (SMF 0/1/2 analysis, cleaning, quantisation, swing, humanisation on copies), recognition of the audio environment (DAWs, audio devices, ASIO, plug-in folders, DAW projects) and Universal Resource Discovery of digital libraries (MIDI, samples, loops, impulse responses, presets, soundbanks, instruments, plug-ins) on every drive. It reuses the existing M10 engines and works with any DAW through standard files. The status register marks the module PARTIALLY IMPLEMENTED because several planned parts are missing.

Neural audio generation with non-commercial weights is excluded. The Session Co-Pilot works through standard files (WAV/BWF/AIFF, SMF) and has no native integration with a specific DAW; role and content recognition is heuristic. In the executable, the installed workflow (Inbox watcher, queue, session workspace) is not available: the analysis, MIDI, reference, separation, environment and library commands are.