c-apple-silicon-wav

This is a delta implementation based on the Apple Silicon version in implementations/c-apple-silicon, extended to play arbitrary WAV files.

For hardware assumptions and receiver setup (such as AM radio placement), see implementations/c-apple-silicon/README.md. This README summarizes only the differences.

Changes from c-apple-silicon

Input format

Changed input from .tune to WAV
Run format: ./main input.wav

WAV loader

Added RIFF/WAVE validation
Scans and reads fmt / data chunks
Handles chunk padding (even-byte alignment)
Downmixes multi-channel audio to mono
Applies input sample rate to playback timing

Audio preprocessing

32-bit-safe peak normalization
Voice-oriented pre-EQ (HP/LP + presence)
Compression and loudness boost via AGC + limiter

Modulation method

Replaced random-sampling-based density generation with a Sigma-Delta (error-accumulation) method
Improves intelligibility by reducing fine-grained random on/off artifacts

Packet handling

Keeps sample time continuity across packets
Sets default PACKET_MS to 100ms
Adjustable at runtime via SBR_PACKET_MS

Supported WAV

RIFF/WAVE
PCM 16-bit (audio_format=1, bits_per_sample=16)
1 channel or more (internally downmixed to mono)
Any sample rate (tracked by internal timing)

Not supported:

float PCM
24/32-bit PCM
WAV with compressed codecs

Build

bash

cd /Users/cho45/tmp/system-bus-radio/implementations/c-apple-silicon-wav
make

Run

bash

./main input.wav

Examples:

bash

./main hello.wav
./main sweep.wav

Main runtime parameters

SBR_DENSITY_EXP: Exponent of the modulation curve
SBR_DENSITY_DEPTH: Modulation depth
SBR_AGC_TARGET: Target envelope level for AGC
SBR_AGC_MAKEUP: Makeup gain after AGC
SBR_AGC_MAX_GAIN: Maximum AGC gain
SBR_PACKET_MS: Packet length (ms)

Tuning guidelines

Still too quiet: increase SBR_AGC_TARGET, SBR_AGC_MAKEUP, and SBR_AGC_MAX_GAIN
Choppy output: increase SBR_PACKET_MS (for example 200 to 500)
Distortion: decrease SBR_AGC_MAKEUP or SBR_DENSITY_DEPTH