new·The score now tells you which way it movedA brain's exam only ever grows: its own material writes questions, and so does every question a real caller asked and did not get answered. The score is a percentage over that growing set, so a brain that learned more could post a smaller number — and this week three did. One of them answered two MORE questions than the week before and showed eighteen points less. Printed as a single percentage, that reads as decline to a reader and as punishment to anyone who contributes material.all news →
mozg.beta
Sign in

AI Asset Pipeline · all subjects

audio-post-processing

5 notes, read out of this brain and free to use. Each one was extracted from a source and is re-checked against its exam.

mp3 overshoots on decode by up to 30%

mp3 files can overshoot peak level by up to 30% when decoded (limiting to 0.891 pre-encode still decoded at 1.117). Check clipping with `post.decoded_peak()` and re-encode at lower levels via `process()` until it fits within limits.

silenceremove at -50 dB destroys quiet cues

Using `silenceremove` at -50 dB destroys quiet cues to nothing (e.g., scatter-land.mp3 at 0.144 s became inaudible). The `process()` function redoes trimming without silenceremove if a trim destroys more than 65% of a clip.

Cosmetic fades on loop=True beds dip amplitude every pass

Cosmetic fades applied to looping audio beds cause the amplitude to dip on every playback loop. Fades are dropped for `loop=True` cues.

process(loop=True) does not set length automatically

`process(loop=True)` only fades edges and does not set the audio length, causing files like a 1.6 s whir to ship as a 10 s file. Use `post.make_seamless_loop(length_s=…)` to explicitly set the desired loop length.

mp3 under 3 LAME frames has no derivable duration

mp3 files under approximately 3 LAME frames (very short clips like 54 ms) have no derivable duration, causing `duration_s()` to return 0.0 and abort a build even if the cue is valid. Fallbacks are in place to handle this.

Give your agent this brain