Skip to content

7.2.0

Choose a tag to compare

@juzzlin juzzlin released this 29 Aug 16:59
· 51 commits to master since this release

Release date:

Sat, 29 Aug 2026 13:16:15 +0300

New features:

  • Add the Speech device

    • Formant speech synthesizer that says a typed phrase, pitched to the notes
      played. English text is read by a letter-to-sound rule set with stress;
      slashes take literal phonemes and an apostrophe marks a stressed syllable
    • A note speaks either the whole phrase or the next syllable, running free, fit
      to a length or one syllable per division
    • Male and female voice types, with Formant Shift, Glide, Sibilance,
      Breathiness, Intonation and Velocity Sensitivity
  • Change the device in a slot without losing the mix around it

    • The Device Rack's Manage menu has a Change Device... item. The new instrument
      takes over the old one's insert effects, sends, fader, gain, pan, fader
      position and send tap, and keeps the slot's place in a Sub Mixer
  • Export, import and copy whole effect racks

    • The Insert Effects and Master Effect Rack dialogs write and read a whole rack
      as a .nahdrack file. A single effect keeps its own .nahdeff
    • Copy Rack... takes the whole chain from another device or from either master
      rack, bypass state included
  • Add the String & Voice V2 device

    • Its Strings section has the switches the hardware has, Upper and Lower, saying
      which half of the split it sounds on, with Balance setting how loud. The 8' and
      4' levels V1 carries are not something a VC340 has: a recording of one confirms
      it, every note's harmonics landing on a single fixed-frequency curve with no
      octave register anywhere in it
    • Adds a Tone control over the Strings section, fitted to that recording. At the
      middle of its travel, where the hardware's was measured, the two agree within
      about 1 dB from 1 to 8 kHz
    • The strings source is a saw with its fundamental lifted about 6 dB, which is
      where the hardware's sits
    • Adds a Velocity Sensitivity control, half by default, over both sections at
      once, as the hardware has one keyboard
    • Voices are summed rather than scaled by how many are sounding. V1 divides every
      voice by the square root of the voice count, which holds the total steady but
      ties each note's level to how many others are held: releasing two notes of a
      chord lifts the one still down by about 5 dB
    • V1 is untouched, so existing songs sound exactly as they did
  • Add Interpolate pan to a selection, and to a track

    • Selection > Interpolate pan matches Interpolate velocity, and the track menu
      now reaches the pan interpolation that was already implemented but had no way
      of being asked for
  • Let a pan value be taken off a note again

    • Delete on the pan column clears the pan rather than writing 0, which is hard
      left and went on overriding the column's own pan. A note carrying no pan
      leaves the column where it was
  • Give the synth's filter sweep somewhere to go

    • HPF Cutoff is now an LFO and Mod EG target, so the high pass can be swept
      alongside the low pass. Moving both is what carries a band up through the
      spectrum instead of only opening the top of one, which is what a sweep needs
    • The Mod EG reaches the same destinations as the LFOs: Pitch, Shape, Volume,
      Resonance and Pan on top of what it had
    • Both target lists are append-only, so existing projects keep their routing
  • Add Song > Overview: a signal-flow map of the project

    • Every device with its insert chain drawn in the order the engine actually runs
      it, so a device whose fader sits after its inserts reads differently from one
      whose fader comes first
    • Sends leave from the tap they are taken at, labelled pre- or post-fader
    • A device claimed by a Sub Mixer is shown routed through it, with its sends
      still live, which is what the engine does
    • Click a box to open its editor
  • Add Piano Synth V3, a corrected revision of the CP-80 model

    • The pickup's sign is kept: a piezo at a rigid termination senses force, whose
      partials alternate in sign, and the strike comb keeps its sign too. Levels are
      untouched; what changes is the shape of the strike
    • A partial standing in for a unison pair now carries the pair's weight, removing
      a six decibel step in the middle of the series
    • The strings come in twos, never threes, and single stringing reaches to F#1:
      a CP-80 is bichord over its whole compass
    • A separate device, so songs written against V2 keep playing V2
    • A Velocity Sensitivity control, defaulting to half: full is the reference's own
      velocity-squared law, lower lifts the softer strikes toward the loud ones
      without changing their tone, so the top of the keyboard is playable without
      hammering it
  • Add the Gain rack effect: a plain +/-24 dB trim with a latching clip indicator,
    for staging level without the colour a limiter or a drive brings

  • Add the Monitor rack effect

    • Stereo, Mono, Left, Right and Side fold-downs for checking a mix by ear,
      insertable anywhere an insert rack is
    • Monitoring only: the selected mode is left out of an audio render

Bug fixes:

  • Let the delay be as loud as it is long: feedback was scaling the signal on its
    way into the delay line as well as round it, so the first echo arrived at
    feedback times the input. At the default 0.3 the delay started 10 dB down, and a
    loud single repeat could not be set at all

    • Feedback now governs only what comes round again, so the first echo arrives at
      the level it went in at and each one after it is a further feedback down
    • Existing projects will hear their delays louder and fuller, the more so the
      lower the feedback. This includes the synth's own delay, which is the same
      effect
  • Stop send effects from cancelling the source: a send bus returns what the effect
    added, measured as the difference against the signal it was handed, and an effect
    whose Mix fades the dry out turned that difference into an inverted copy of the
    source. Delay and Chorus had it worst, their wet already being at full halfway up
    the control, so the whole top half of the travel only thinned what it was sent

    • On a send every Mix law now blends additively: the dry passes through whole and
      the wet is added on top, so the entire travel adds effect
    • Fixed for Delay, Chorus, Auto Filter, Phaser, Stereo Enhancer, Stereo Exciter,
      Wave Designer, Analog Fuzz, Bass Grinder, Drive, Saturator and Tube Stage
    • Output gain no longer reaches the dry on a send either, which had been leaving a
      scaled copy of the source in the return at any Mix, including none
    • Insert racks are unchanged
  • Give Dual mode's Voice Depth something to hear: the pair opened to only 20 cents
    at full depth, a fifth of what every other detuned mode reaches, so the knob had
    nowhere to go

    • Both halves now land on the ends of the same spread Unison opens up, 100 cents
      apart at full depth. Fixed in the wavetable synth too
  • Fix the clicks a swept filter left on every note: a filter parked at the end of
    its range passes the signal through, but it was being skipped rather than kept
    running, so a sweep coming back re-entered it with state from tens of
    milliseconds earlier and stepped the output

    • The handover is crossfaded now, so the moment the sweep crosses the threshold
      is continuous instead of a switch
    • A filter left parked still passes the signal through untouched, so a patch
      that never sweeps sounds as it did
  • Make an audio render reproducible: the same project rendered twice now gives the
    same file

    • Every generator is re-seeded and every free-running oscillator put back to its
      starting phase when a render begins, instead of carrying on from wherever the
      session's earlier playback left them
    • Devices are summed in a fixed order while rendering; which worker happened to
      pick one up was changing the result, float addition not being associative
  • Fix the selection covering the wrong columns after a column was inserted or
    deleted: a column's index is its identity, not its place on screen, so the
    selection now spans the columns actually dragged across

    • Velocity interpolation over a selection was reaching the wrong columns too
  • Sort the effects gallery alphabetically: the list had drifted out of order as
    effects were added over the years

  • Give the internal Volume component its own type ID: it shared the Panner's, so a
    serialized Volume would have come back as a Panner

  • Disable the synth's Roundness knob on every waveform but the pulse, which is
    the only one it affects

Other:

  • Call the control that scales a note's level by how hard it is struck "Velocity
    Sensitivity" on every device. The Synth said "Vel Sens", String Ensemble and
    String & Voice V2 just "Velocity" and "Sensitivity"