Stereo

Stereo signal operations. Akkado tracks channel count (Mono vs Stereo) on every signal; every audio DSP op is stereo-native — one instruction handles both channels with independent per-channel state, and a mono input widens automatically, so you don’t have to duplicate a chain for L and R.

stereo

Create a stereo signal from mono or two separate signals.

Param	Type	Description
L	signal	Mono signal (duplicated to both channels), or left channel if `R` given
R	signal	Right channel (optional)

// Duplicate a mono signal into stereo
saw(220) |> stereo() |> out(@)

// Build a stereo image from two oscillators
stereo(saw(218), saw(222)) |> out(@)

mono

Sum-to-mono downmix of a stereo signal: out = (L + R) * 0.5.

Param	Type	Description
sig	stereo	Stereo signal to downmix

Standard sum-to-mono convention: correlated content stays at 0 dB; uncorrelated content sums at roughly -3 dB RMS.

// Downmix a stereo reverb tail before sidechain detection
wet = src |> stereo() |> freeverb(@, 0.8, 0.5)
env = wet |> mono() |> env_follower(@)

Calling mono() on a mono signal is a compile-time error; call it only on stereo signals. (When passed a function instead of a signal, mono() falls back to the monophonic voice-manager form: mono(instrument).)

left, right

Channel extraction from a stereo signal.

s = stereo(sine(220), sine(330))
left(s)   // → Mono
right(s)  // → Mono

pan

Panning with equal-power (constant-power) law.

Signature	Behaviour
`pan(mono, pos)`	Position a mono signal in the stereo field. Emits `PAN`.
`pan(stereo, pos)`	Balance an already-stereo signal. Emits `PAN_STEREO`.

Position is -1 (hard left) … 0 (centre, -3 dB) … +1 (hard right).

// Mono → stereo pan
saw(220) |> pan(@, lfo(0.25)) |> out(@)

// Stereo balance on a stereo bus
stereo(saw(218), saw(222))
  |> pan(@, 0.3)
  |> out(@)

The stereo form is DAW-style balance, not re-panning each channel. At pos = -1, the right channel is silenced and the left passes through at unity; at pos = 0, both channels are scaled by cos(π/4) ≈ 0.707.

width

Stereo width control using mid/side processing internally.

// Narrow to mono (width=0), original (1), wide (>1)
stereo_sig |> width(@, 1.4) |> out(@)

ms_encode, ms_decode

Mid/side processing: encode stereo to M/S, process, decode back.

// M/S processing: boost sides (wider stereo) while preserving centre
stereo_sig
  |> ms_encode(@)
  |> @ * stereo(1.0, 2.0)   // scale mid 1.0, side 2.0
  |> ms_decode(@)
  |> out(@)

pingpong

True stereo ping-pong delay where echoes cross between L and R.

Convenience: pingpong(stereo, time, fb, width?) Explicit: pingpong(L, R, time, fb, width?)

Param	Type	Default	Description
sig	stereo	—	Stereo input (or explicit L, R)
time	signal	—	Delay time (seconds)
fb	signal	—	Feedback (0..1)
width	signal	1.0	Pan width (0 = centre, 1 = full ping-pong). Optional.
dry	signal	1.0	Dry-mix coefficient (Category A). Kwarg only.
wet	signal	0.5	Wet-mix coefficient (Category A). Kwarg only.

The output applies the unified mix line per channel: out = dry_in * dry + delayed * wet. Category A defaults give a balanced parallel mix where the echo sits 6 dB below the dry signal. Pass wet: 1.0 to recover the pre-0.5.0 behaviour where the wet signal was added at full level on top of the dry source.

saw(110)
  |> stereo()
  |> pingpong(@, 0.375, 0.6)            // defaults: dry=1, wet=0.5
  |> out(@)

saw(110)
  |> stereo()
  |> pingpong(@, 0.375, 0.6, wet: 1.0)  // legacy-loud echo tail
  |> out(@)

Stereo-native effects

Every audio DSP op (lp, hp, delay, freeverb, saturate, …) is stereo-native: the compiler emits one instruction flagged STEREO_OUTPUT, and the opcode handles both channels in a single dispatch with independent per-channel DSP state. A mono input widens automatically — no stereo() wrapper needed.

// One instruction per effect; each handles L and R in one dispatch.
saw(220)               // mono — widens automatically
    |> lp(@, 500, 0.7)        // stereo lowpass, per-channel state
    |> delay(@, 0.25, 0.5)    // stereo delay lines per channel
    |> freeverb(@, 0.85, 0.5) // stereo reverb with cross-coupling
    |> out(@)

Scalar / control-rate parameters (cutoff, Q, feedback, …) are shared between the L and R channels. If you want independent parameters per channel, split the chain manually with stereo(lp(left(s), cutL), lp(right(s), cutR)).