Skip to content

Dodaios is an open-source agent OS for your desktop: it runs your coding agents, browser and apps by voice or chat, on your own machine. Formerly Personal Jarvis.

feat(appshot): make the capture cue a crisp click-clack

rubenluetke10-beepdb40d42224d1
diff --git a/jarvis/audio/chime.py b/jarvis/audio/chime.py
index e41ac958dd..91b3d15479 100644
--- a/jarvis/audio/chime.py
+++ b/jarvis/audio/chime.py
@@ -93,58 +93,39 @@ SCREEN_CAPTURE_PREROLL_S: float = 0.045
def generate_screen_capture_pcm(
sample_rate: int = 24_000,
- amplitude: float = 0.2,
+ amplitude: float = 0.22,
) -> bytes:
- """Generate the appshot cue: a soft, rounded two-note "bloop".
+ """Generate the appshot cue: a crisp, tactile "click-clack".
- A sine that glides up a fourth (D5 -> G5) lands on a gentle D6 bell, both
- with a smooth attack and a short synthetic room tail. Every onset sits
- after ``SCREEN_CAPTURE_PREROLL_S`` of silence, so a freshly opened stream
- plays the whole cue (the former noise-burst shutter packed half its
- energy into the first 5 ms and was inaudible). Synthesized in memory:
- portable and free of third-party recordings.
+ Two short clicks, like a phone camera shutter. Each one is a damped high
+ tick (a few kHz) over a whisper of band-limited noise, with a 1 ms attack
+ so it snaps without a digital edge; the second is a touch lower and
+ softer. Every onset sits after ``SCREEN_CAPTURE_PREROLL_S`` of silence,
+ because a freshly opened stream fades and swallows its first milliseconds.
+ Synthesized in memory: portable and free of third-party recordings.
The peak stays low on purpose: the player's master volume adds up to 4x
- makeup gain, and a louder source would ride the peak limiter instead of
- staying soft.
+ makeup gain, and a louder source would ride the peak limiter.
"""
preroll = int(SCREEN_CAPTURE_PREROLL_S * sample_rate)
- body_s = 0.34
+ body_s = 0.12
n = int(body_s * sample_rate)
t = np.arange(n, dtype=np.float64) / float(sample_rate)
+ rng = np.random.default_rng(0x5C4E_454E)
+ # Differentiated noise tilts the spectrum up: an airy tick, not a hiss.
+ noise = np.diff(rng.uniform(-1.0, 1.0, n + 1))
- def _note(
- onset: float,
- f_start: float,
- f_end: float,
- glide_s: float,
- attack_s: float,
- decay: float,
- ) -> np.ndarray:
+ def _click(onset: float, pitch: float, strength: float) -> np.ndarray:
local_t = np.maximum(t - onset, 0.0)
active = t >= onset
- # Smoothstep glide, integrated into phase so the pitch bend has no seam.
- g = np.clip(local_t / glide_s, 0.0, 1.0) if glide_s > 0 else np.ones_like(t)
- freq = f_start + (f_end - f_start) * (g * g * (3.0 - 2.0 * g))
- phase = 2.0 * np.pi * np.cumsum(freq * active) / sample_rate
- tone = np.sin(phase) + 0.12 * np.sin(2.0 * phase) # a touch of warmth
- attack = 0.5 - 0.5 * np.cos(np.pi * np.clip(local_t / attack_s, 0.0, 1.0))
- return tone * attack * np.exp(-local_t * decay) * active
-
- dry = _note(0.0, 587.33, 783.99, 0.05, 0.008, 20.0)
- dry += 0.8 * _note(0.075, 1174.66, 1174.66, 0.0, 0.010, 13.0)
-
- # A few soft, darkened early reflections give the cue air without a
- # convolution dependency.
- wave = dry.copy()
- for delay_s, gain in ((0.031, 0.22), (0.053, 0.15), (0.083, 0.09)):
- shift = int(delay_s * sample_rate)
- echo = np.zeros_like(dry)
- echo[shift:] = dry[:-shift]
- echo = np.convolve(echo, np.ones(6) / 6.0, mode="same")
- wave += gain * echo
-
- fade_out = np.clip((body_s - t) / 0.04, 0.0, 1.0)
+ attack = np.clip(local_t / 0.001, 0.0, 1.0)
+ tick = np.sin(2.0 * np.pi * pitch * local_t) * np.exp(-local_t * 260.0)
+ body = np.sin(2.0 * np.pi * pitch * 0.5 * local_t) * np.exp(-local_t * 140.0)
+ air = noise * np.exp(-local_t * 420.0)
+ return strength * attack * (0.55 * tick + 0.25 * body + 0.35 * air) * active
+
+ wave = _click(0.0, 3400.0, 1.0) + _click(0.048, 2600.0, 0.7)
+ fade_out = np.clip((body_s - t) / 0.01, 0.0, 1.0)
wave *= fade_out
wave *= amplitude / max(float(np.max(np.abs(wave))), 1e-9)
wave = np.concatenate([np.zeros(preroll), wave])
diff --git a/tests/unit/audio/test_chime_ready.py b/tests/unit/audio/test_chime_ready.py
index dced9153f4..9800428211 100644
--- a/tests/unit/audio/test_chime_ready.py
+++ b/tests/unit/audio/test_chime_ready.py
@@ -47,15 +47,13 @@ def test_screen_capture_cue_is_original_and_deterministic() -> None:
def test_screen_capture_cue_survives_a_fresh_stream_onset() -> None:
- """The cue must not live in its first milliseconds.
+ """The cue must start after a silent pre-roll.
``play_pcm`` fades the first 5 ms and a freshly opened device swallows its
- start; the old shutter packed ~half its energy there and was inaudible.
+ start; a click that begins at sample 0 loses its whole transient there.
"""
samples = np.frombuffer(SCREEN_CAPTURE_PCM, dtype=np.int16).astype(np.float64)
preroll = int(SCREEN_CAPTURE_PREROLL_S * CHIME_SAMPLE_RATE)
assert not samples[:preroll].any()
- energy = np.cumsum(samples**2)
- early = energy[preroll + int(0.005 * CHIME_SAMPLE_RATE)] / energy[-1]
- assert early < 0.05
+ assert np.abs(samples[preroll:]).max() > 0
assert np.max(np.abs(samples)) < 32767 * 0.5 # soft, never clipping