update documentation.html for 1/24/2025 gradio/massive overhaul

fastrtc issues solved mostly in gradio app. html web tester still better ux for the moment

we keep making svelte angry

MOAR LOGGING

one more try

one more try

vibe logging rtc config type error issues

it wasn't the mount path's fault... trying turn credentials async again with logging

it wasn't the mount path's fault... trying turn credentials async again with logging

lets try mounting gradio at root like we should...but only in serve mode for the moment

one more try at this turn credentials stuff before we just mount gradio at root instead

different rtc_configuration using turn

rtc_config in stream handler to make hf happy

i think we did it? gradio with fastrtc v1

hmm...reverting fastrtc stuff while we test locally

maybe we can do this in gradio...

maybe we can do this in gradio...

maybe we can do this in gradio...

updating web tester and simplifying websockets route

massive rewrite...pink noise warmup, no more crossfading needed, one_shot_generation fixes, jam_worker overhaul...still gotta fix html web tester now due to system.py changes in mrt

buffer queue customization added to websockets route

realtime mode queue increase

claude hallucinated a jax version

solving for those nightlys that no longer exist lol

manual reset for /generate. something's still accumulating...

more cleanup in one_shot_generation to avoid stale context

one shot generations start failing after a few successful ones...

much closer now with the loudness match inside ableton

remove peak normalize from one_shot_generation

loudness is always a puzzle in the DAW...one more attempt

pyloudnorm is yelling at us

reintroducing better loudness matching in /generate

braindead progress updates for /generate to be used in JUCE

fixing bar-aligned context inside /generate route just like we did for jam_worker

reverting loudness matching for /generate

bad import in one_shot_generation.py

loudness matching improvement for single-shot generations in DAW

smarter loudness matching in /jam/start

yet another _append_model_chunk_and_spool rewrite to fix silence gaps

reverted

sometimes a claude yolo pt2

sometimes a claude yolo

i declare debugging war

it wasn't the loudness matching. _append_model_chunk_and_spool debug now

fixing logging

loudness matching debug logs for sanity

one more attempt at _append_model_chunk_and_spool

reverting again...we keep making it worse

another _append_model_chunk_and_spool revision

another failure at getting rid of these silence dips. reverting