Complete Crea Video Lyrics guide

First the MIDI, then the audio chain

These are two different, complementary processes. MIDI corrections change musical data before synthesis; the audio chain processes the generated sound. Here is how to find your way:

1. MIDI performance: instrument-family corrections, legato, articulation, vibrato and expression, repeated-note handling and groove for mechanical channels. Select corrections to suit the song; the original stays unchanged. Instrument corrections.
2. Synthesis and sounds: choose the engine and SoundFonts, including per-channel choices, and adjust the balance. The sound bank and the source MIDI still matter. Engines and adjustments.
3. Audio processing: EQ and Presence, “Mix glue” compression, stereo control and IR convolution reverb, plus valve warmth and multiband bass options; the Options panel also includes Balance volumes, Protect the vocal and ensemble-instrument doubling. The mixer measures loudness and peaks to manage the final level; you do not need every effect enabled. Options panel.
4. Style production: the profile measures the mix's frequency balance, stereo and dynamics and calculates corrections towards the selected genre's ranges, with adjustable intensity. This audio chain runs locally and does not require cloud AI. Style Selection.
5. Files and video: preview and export the MP3/WAV backing track, then move to Syncro Lyrics and Crea Video for lyrics and MP4. If the MIDI has lyrics, the program can take you straight to Crea Video when the export ends, with the MP3 and SRT already filled in. Corrected MIDI export saves musical data instead: audio effects do not become part of the MIDI file. Export corrected MIDI.

English guide: Start screen · Midi to Mp3 · Syncro Lyrics · Crea Video

Resume work from the song archives

• MIDI song archive: reopens the MIDI and its settings in Midi to Mp3.
• MP3 song archive: finds matching MP3 + TXT files and fills the Sync lyrics fields to prepare an SRT.
• Video song archive: finds matching MP3 + SRT files and fills the Create video fields; then choose a background. It is not a list of finished MP4 files.

Scanning includes subfolders and replaces the entire previous list of the selected archive. Cancelling preserves that list; a completed scan with no matches clears it. Removing an entry never deletes the files. Create video remembers your font, size and last background folder after restarting. Each module has its own Songs menu (and ? with Help and About): the MP3 archive opens from Sync lyrics, the Video archive from Create video.

Archive details · MP3 and text for synchronization · Video sources

SRT/.parole matching accepts consecutive joined lines and differences in spacing, case and punctuation while preserving original timing.

Compensated stereo level: widening and narrowing preserve the mix’s average spectral balance. The program measures mid and side energy in windows distributed across the song and computes a frequency-dependent correction held constant throughout the song: it reduces boosts and recovers losses caused by the stereo effect alone, within 12 dB. It applies no fixed cymbal cut and changes no MIDI notes, velocities or drum kits. Intensity 1.00 bypasses this stage. Calibration uses the accompaniment even when the guide is audible, for preview and audio export with engines 0, 1 and 2. An additional measurement and audio pass are required; local balances may differ from the song average.

Humanization and pedal: added note-release extensions cannot cross sustain, sostenuto, reset or note-stop events. This prevents an already released note from being captured by a later pedal press. Original pitches, onsets, controllers and authored overlaps are preserved; controllers on other tracks of the same channel are included.

Contextual crash and ride balance: with Balance volumes enabled, preview and audio export in engines 0, 1 and 2 measure SF2 crash and ride separately, including mixed SFZ arrangements. Attacks are compared with instruments playing at the same time; the correction remains constant throughout the song. A tolerance band avoids unnecessary changes: up to 6 dB attenuation or 2 dB recovery, with no recovery for parts written entirely as soft hits. Kick, snare, hi-hat and other drums retain the kit balance. Manual levels remain applied and are excluded from the automatic decision. Enough hits and at least two reference instruments are required; insufficient evidence or an inaccurate kit reconstruction bypasses the control, with a diagnostic log message. The comparison band is a conservative estimate requiring listening checks. Three additional drum renders and measurements are needed. Original MIDI files remain unchanged; this does not affect MIDI-only export or SFZ drum kits.

Preview files: intermediate mix channels are removed after preparation, including errors or Stop, and use names separate from export channels. A solo preview WAV is removed when playback ends or Stop is pressed. The completed full mix remains available for the cache. Exported channel WAVs are also removed after the MP3 is created.

M-Live lyrics: in addition to standard lyric and karaoke events, the reader recognizes M-Live SysEx text lines with dedicated syllable progress controllers. A valid original clock produces SRT, .parole and intro files without interpolated word timing. Once both MP3 and SRT exist, the export offers to open Create video. Missing usable lyrics or a write failure is explained in the completion message. Corrected MIDI export preserves the karaoke clock even when its data channel plays no notes. Older corrected files that already lost these controllers must be regenerated from the original.

Disk space after export: after the MP3 is successfully created, the song’s individual channel WAVs are removed automatically, including leftovers from previous exports of the same song. They are not a permanent cache. Temporary channels created by the current operation are also cleaned up on Stop or error. MP3, MIDI, SRT, .parole, intro and the completed full preview cache remain available. To change the sound after export, adjust the settings and run Create Audio File and SRT for Video again. Family SF, Cancel and Save and Apply are centered as a group in the panel.

Contextual drum and wind balancing: the average level over the entire song no longer triggers an extra drum cut on its own: dominance must be confirmed against instruments playing at the same time. Wind-section energy is compared with at least two active accompaniment parts: attenuation is limited to 3 dB and requires persistent excess (at least 3 consecutive seconds and 60% of the channel's activity). Brief responses, solos without enough references and winds already behind the section remain unchanged; no fixed boosts are applied. The guide, muted parts, manual levels and timbral compensation are excluded from the decision. SFZ descriptions containing octave-compensation text are also recognised. Applies to balanced preview and audio export in engines 0/1/2; original MIDI files and the approved vocal-protection rules are unchanged. This conservative estimate requires listening checks.

Contextual recovery of recessed parts: with Balance volumes enabled, after audio balancing a recessed part is compared with the original MIDI rendered using the same bank, engine and actual parameters. Recovery requires a measured processing loss and at least two simultaneous reference instruments: up to 3 dB, limited by both the original level and accompaniment headroom, with gradual attack and release. Only persistent passages outside the vocal are eligible; the guide and a half-second guard, deliberately soft notes, ramps, drums, pads and fret noises are excluded. Naturally quiet SoundFonts are not automatically boosted and selected timbres are not replaced. Incomparable arrangements or presets are skipped with a diagnostic reason. Preview and audio export, including SFZ rendering, may need extra measurements and rendering of candidate reference parts. Original MIDI files and manual levels are preserved; listening checks remain necessary.