Skip to content

Latest commit

 

History

History
198 lines (190 loc) · 8.81 KB

todo.md

File metadata and controls

198 lines (190 loc) · 8.81 KB
Raw
  • Finish sprae 10
  • no-caret (0 focus) play bug
  • space repeat
  • head of audio is boosted for some reason
  • [~] discrepancy of caret with sound -> can't reproduce
  • click on time must not reload anything
    • loading link with time should navigate to the line

Backlog

  • Automatic tests
  • Shift play button to the left
    • Display current time
  • All editing operations
    • delete
    • ctrl+C / ctrl+V
  • Undo/redo history (separate from browser history)
  • Save result
  • Adjustable view
    • block size
    • color theme
  • player backends
    • Audio
    • WAA
  • Drag-n-drop
  • Separate by fragments (scenes) via enter
  • Playback bar with current time, play/stop, more
    • Position: bottom floating/appearing, bottom fixed, balloon next to cursor, no (melded into UI)
  • Operations
    • Normalize audio (from playback bar?)
    • Revolume selected fragments
    • Noise-gate plugin
    • Speedup silences (plugin?)
  • Change number of channels
  • 11labs integration: generate speech of length
  • Switchable main-thread / worker / GPU processing

Reiterating

  • display audio
  • play audio
  • caret indication
    • Don't update caret in raf: update only on playback and time change
    • Don't track caret on every focus: only when user selects by mouse
    • Make playback within the selection
  • Fix safari
  • loses caret on play, like insert silence, press play etc
  • serialize file in url: ?src=path/to/url/file/to/fetch
  • sprae :onfile-attachment-accepted
  • add preloader (sprae mount-unmount)
  • delete fragments -> updates audio
  • create silence by space
  • download
  • caret must be able to be reoriented during the playback
  • Safari: wrong current time positioning
  • BUG: stopping drops focus
  • Make 'Enter' create segments
  • time codes next to lines
  • br
  • del
    • fix deleting tail properly
  • normalize
  • BUG: setting caret to the beginning of segment (a bit from the left of segment) doesn't start playback properly
  • faster encoder by just copying changed subbuffer data, opposed to full rerender
  • fix playback multiple segments
  • Add vertical shift of average
  • Shift + select
  • interleaved buffers pointing to chunks of wav file, rather than audiobuffers same as below
  • immediate audio ops via copy - saves 15ms, takes a lot in terms of losing AudioBuffer primitive
  • worker processor
  • actions via beforeinput inputType

[ ] MVP: basic dubs editor

  • Make delete: from-to signature instead of from-count

  • Debounce delete better

  • "Open audio"

    • "Generate speech" or "Pick random audio" intro screen. (+ button at the right) -> use more complete sources config

  • Reflect operations in URL

  • Backspace-deleting from the beginning of segment doesn't remove break but deletes tail of prev segment instead

    • join operation that serializes as removing break

  • mute

  • take source from URL.

    • if there's none - take random source

  • support dropping files

    • save dropped files to storage

  • Make history of changes with undo/redo

  • Time-codes of following segments are messed up: make them href-able

  • Bug: insert silence at the beginning of new segment -> feature

  • Save local file edits to kv-storage - saved in history

  • BUG: 0:60 in timing

  • OPTIMIZATION: use onbeforeinput/oninput for handling operations

  • BUG: deleting from left & then from right of caret is different

  • BUG: fix playback from caret

  • Alt-Space for start/stop

  • Loop play selection

  • Outsource audio-decode, add missing codecs

  • Outsource media loopStart / loopEnd

  • Better selection logic: must be immediate

  • Display open/loading status

  • Display + for newlines

  • . for silence

  • Empty URL shows "Open file"

  • Loads source from url on init

  • [?] Display loading status in playback not sure still if we need playback

  • Show average line in samples use dots instead

  • Deleting, changing caret, deleting again causes UI waveform assertion fail

  • Small screens wrongly wrap waveform timing

  • Deleting part of audio screws up play button position

  • End of file caret positioning is wrong

  • Delete-all case doesn't get saved

  • Big file editing generates tons of error logs : must be good

  • Big files break caret line at the end (see bvg)

  • Loaded file misses offset

  • Played waveform update on big files is very slow. Use overlap technique or virtual list via intersection observer

  • Safari: smooth audio currentTime (opposed to glitchy now)

  • A way to download / reverse / etc selected fragment (... at the right)

  • Stopping playback causes glitch

  • Bug with assets/1s.wav playback - end line caret shifts down

  • Problematic mobile rendering

  • Mobile playback doesn't start from selection

  • Bug: renavigating during play

  • Bug: mobile safari play button sticks glitchly (alternative to intersection observer?)

  • Bug: multiline selection is damaged

  • Bug: empty head starts playing something non-silence

  • Bug: needs enhanced lines calculation, ideally chars-per-line

  • Bug: playback with space is glitchy (resets caret)

  • Bug: deleting is broken

  • Bug: doesn't renavigate by click

  • Bug: doesn't scroll on caret offset

  • Bug: loop playback selection is broken

  • Make play always cover the time, then it leaves space for "record" button

  • Zoom

  • Render only visible part (virtual) - must reduce rendering load significantly

  • Resize throttle

  • <time-codes>

  • <playback-panel>

  • <wave-area>

  • Autoplay, loop, current line - can be navigatable from URL

  • Add info icon: support, github, brahman

    • tips, generate theme
    • adjust settings: audio loudness metric, block size

  • support paste fragment of itself

  • Mark loop selection

  • Mark fragments

  • Detect characters per line via ranges method: https://www.bennadel.com/blog/4310-detecting-rendered-line-breaks-in-a-text-node-in-javascript.htm

  • Separate audio-decode module with all codecs...

  • Make play button clickable area big

  • Recent files

  • use media-offset for looping -> own function play-loop

  • detect cmd/ctrl key depending on platform

  • make player responsive in mobile as bottom play button with overlay

  • make playback sticky to avoid hiding playback (intersection observer + position change)

  • use plain (interleaved?) arrays instead of audio buffers - faster decoding, faster transfering to worker - limited maintainability, no need to transfer to worker

  • use decodeAudioData main thread "worker" for faster decode, detect supported native codecs & video

  • Loudness weighting

  • Better loudness display: it is inadequate now

  • Display left/right channels with half-transparent blacks, and black is their intersection

  • time codes as # hrefs

    • make navigatable

  • Faster updates: maybe no point rerendering/encoding full waveform, or parallelize, or faster wav encoder (wasm?)

    • ? should we work straight on wav buffer maybe instead of audio buffers?

  • Highlight of playable/playing region via diff color

  • use audio-buffer-list for faster ops? -> use own implementation

  • theme selector: color gradientish, inverse, cool, hot, dynamic, bw, font style

  • move loading/decoding/encoding to worker

  • random phrase player (from URL - like thetamath) via free speech api

  • broken sprae condition of :if :ref

  • convert ops units to h/s/ms

  • replace file selector with + under caret?

  • open file

  • Make nicer playback UI (bottom of page player) -> not proved to be the best

    • Errors and loading state must be indicated there
    • Precise current playback time

  • delete file fully -> displays open file again

  • save file in storage? -> can be done via browser caching

  • Safari: initial audio loading state displays Error (show silent buffer)

  • scroll must follow the current caret position

  • save edits in URL, so that any audio URL can be opened, edited, played.

  • More audio transforms

  • Make reusable (web-) component

    • Textarea mode: no-line-breaks simple renderer on any textarea, no autosizer
    • Adjustable timecodes
    • Adjustable menu
    • Adjustable theme
    • Adjustable line breaks / ops

  • Think of embeddable links

  • Recording capability

  • Add tests (playwright?)

  • Measure via LUFS and other methods

  • Process audio with lino?

  • Vary color based on spectrum

  • ?Use timing object https://github.com/chrisguttandin/timing-object -> nah

  • Editable labeling / phrases