The sound editor handles microphone input and stored tracks. A shared sound manager handles playback streams. The research identifies the scene, stream creation and cleanup, ADPCM routines and microphone calls in the studied builds. After a recording request, follow the remaining steps to check when the encoded audio is actually saved into the PPM. [1]
Capture path
Microphone samples enter a temporary capture path, are prepared for the selected track and only later become serialised audio. The recorded track length must describe the encoded bytes that will actually be written. Check both the temporary recording and the encoded track when audio works in the editor but fails after saving. [1][2]
Playing and stopping tracks
Background music and the three effects have different playback roles. Effects are started by per-frame flags; BGM is a continuous stream with its own speed. Seeking, stopping and restarting a note must cancel or reposition the old stream before creating the new one, otherwise a previous sound can continue into the next loop. [1][2]
What has been checked
The mapped functions establish parts of Studio's recording and playback control flow, not every hardware timing detail or microphone sample condition on all consoles. When comparing an emulator, hardware and a browser implementation, capture the sample rate, codec input and speed fields alongside the audible result. [1]
Inside the V2 sound editor
The mapped SoundEditor scene has separate controls for BGM, SE1, SE2, SE3, mix and frame mix. Its input handler checks whether a track contains audio, then handles recording, deletion, preview and exit in different states. A button changing appearance cannot by itself establish that the document audio has changed; the subsequent recording and save path must be followed. [1]
The shared sound manager has eight slots for active sequence handles. A slot retains a handle, archive and sequence identifiers, a short keepalive and fade state. An existing handle can remain after audible playback has ended; periodic updates stop or detach it. If a sound seems to persist after leaving a page, inspect the manager's slot lifecycle as well as the document track. [1]
Recorded sample and codec details
The original US sampling evidence records 8,180 Hz with timer period 4,097. The examined effect-recording routine has an 8,192-byte capacity, and the encoded wave descriptor reserves a 12-byte header before its ADPCM payload. The encoder can write a final low nibble for an odd sample count while reporting only complete output bytes. A test that compares just the reported length can miss a final-byte difference. [1]
Those numbers come from the routines examined in the research. They do not settle exactly where the SDK stops a partially filled microphone buffer, nor do they prove how an existing BGM is replaced after overdubbing. Keep capture, encoding, document assignment and PPM serialisation as separate checkpoints when diagnosing a recording fault. [1][2]