derselbst / anmp Goto Github PK

for crossplatform

via dbus. therefore write a launcher

dont let the user trigger the next song by seeking beyond the end of the current song. just as one cannot seek beyond its beginning. catch that case within GUI?

reproduce:

• play a song
• seek forward
• pause
• play
• seek backward
  --> next song triggered

implement new render technique introduced in 1f6929f for all libs, to obviate #10

read reverb from fludisynths dedicated fx channel, to avoid switching off reverb when muting voice 0

in Player class: provide callback function for playhead and currentSong changed by internal playback-thread

right click a song > properties: open up a nice gui that gives all informations about the selected song:

• metadata
• loop points, i.e. visualize loop tree??
• total playing time (with / without loops?)
• total frames available
• sample format, rate
• which wrapper lib currently handles this file
• edit available voices and their channel config
• add/remove voice muting presets

Figure out why dataChanged in PlaylistModel doesnt cause calling PlaylistModel::data() to refresh cells and their styles and colors:

PlaylistModel.cpp:192

This would avoid resetting the whole model (and by that redrawing the whole view), just because the next song is played.

therefore steal the code from madplay

implement a custom midi player so that individual track can be looped as here:
http://level42.ca/projects/ultra64/Documentation/man/pro-man/pro20/index20.5.html

FluidsynthWrapper becomes LibSMFWrapper, holding an arbitrary singleton of class Synthesizer (or so) which is solely responsible for synthesizing. Fluidsynth is a derivative of this class.

currently only works when rendering whole song active

that occur whenever JackBufSize > FramesToRender && JackSampleRate < SongSampleRate

• delayed start of a song (fixed by 72dd1fc)
• interrupted audio or no playback at all

currently doesnt work at all

use qsplitter and dock widgets

instead of sample peak as soon as libeubr128 got rid of speex dependency

PlaylistFactory: no try and error with modplug, solely it by fileextension

e.g.: for one decode do noise normalization + wave output

when playing next file, probably due to resetting the model

Expression 'res' failed in 'src/hostapi/alsa/pa_linux_alsa.c', line: 3497
Expression 'PaAlsaStreamComponent_EndProcessing( &self->playback, numFrames, &xrun )' failed in 'src/hostapi/alsa/pa_linux_alsa.c', line: 3596
Expression 'PaAlsaStream_EndProcessing( stream, framesGot, &xrun )' failed in 'src/hostapi/alsa/pa_linux_alsa.c', line: 4462
Expression 'res' failed in 'src/hostapi/alsa/pa_linux_alsa.c', line: 3497
Expression 'PaAlsaStreamComponent_EndProcessing( &self->playback, numFrames, &xrun )' failed in 'src/hostapi/alsa/pa_linux_alsa.c', line: 3596
Expression 'PaAlsaStream_EndProcessing( stream, framesGot, &xrun )' failed in 'src/hostapi/alsa/pa_linux_alsa.c', line: 4462

esp. for vgmstream: play song A, play next song B

B may use pcm buffer of A. if vgmstream isnt ready yet, one will hear very first part of song A

--> introduce additional sync variable at VGMStreamWrapper.cpp:67

implement a class PlaylistModel that derives from QAbstractTableModel. This class the songs that are added by the user (i.e. it holds an instance of IPlaylist). It will also be used to display the songs in a tableView, allowing it to edit the playingorder, retrieving information of the song that are in the playlist (e.g. SampleRate, Metadata, FileFormat, Filepath, looppoints, etc.)

TestCases:

• add file to playlist, remove file, play file
• exception while parsing cue sheet

having big wave files (>2.5 GiB) containing pcm as floats, this command may take up to 15 seconds to process. unacceptable! therefore, make the sample format generated by sndfile depend on the sample format of the input files, i.e. if we get an audio file with floats or doubles, output them as floats; if we get a file containing integers, output them as int32.

for playlist model and disable songs if false

split all current public methods into public and private part, private part does actual job, public calls private and syncs with future

like ffplay uses

a preset defines mute masks for voices, enabling easy switch between normal, cave or aquatic theme of a song.

The Song class has to provide several voices, that may be mixed or muted by IAudioOutput. Every voice itself may play on a mono, stereo, surround or whatever n-channel. For simplicity this configuration must be the same for every voice (illegal that one voices plays on mono channel, while another has a stereo channel).

introduced in 0cd5ebd

create a way to configure the channel layout for a song. make sure it doesnt get overwritten by subsequent call to Song::open().

esp. usefull for libgme: when playing spc we should get 8 stereo channels from the DSP. by using bitmask become able to play only specific channels.

best way to do this to use Song::data as two-dim array, like

pcm_t data[Song::Format::Channels][Song::getFrames()]

by that we also get rid of Song::count

in this case dont hold whole song in memory but only Config::FramesToRender frames

https://forum.kde.org/viewtopic.php?f=66&t=121086
https://specifications.freedesktop.org/desktop-entry-spec/desktop-entry-spec-latest.html

... in order to display exceptions

create lamdba and calculate volume in callback function

introduce bool useCached parameter to speed up file openings??

and compare samplerate and format explicitly

use ReplayGain or loudness-scanner to achieve a noise normalization for all audio format (even those, which dont support taking in such loudness-metadata)

possible solution: write loudness info to (hidden) text files