rripio / pre.di.c Goto Github PK

Hello. Some work has been done in my fork to show playing information at the web page bottom.

The new web page depens on:

scripts/server_local.py
bin_custom/server_local.py
bin_custom/players.py

These new programs are under bin_custom/ in my fork, but can be placed under bin/

When launching clients on startup some seems to need some extra time as implemented in 5b7779e, even if its ports are up as reported by predic.wait4source().

mpd is the more connection-reluctant as of now, and needs an extra time of 2 seconds, written as time.sleep(gc.config['command_delay'] * 2)

This added wait interval could be a candidate for a separate config variable, since it is more client dependent than machine dependent.

As a remainder, let's note that gc.config['command_delay'] configures a global coefficient for all hard delays in the program, in the hope that they scale well with machine speed.

Download and update scripts can be adapted from the FIRtro distro to work for pre.di.c. Some work has been done on my fork

Hola Roberto,
No sé si te acuerdas de mí de la época del SweepScope. Ya estoy muy fuera de la acústica :o(.

Me gustaría contactar contigo. Yo no he cambiado mi email (supongo que si buscas en tu correo por SweepScope, lo encontrarás). Si lo tienes ¿podrías escribirme? Yo tengo el tuyo de la época de "wanadoo", que ya no está operativo. Por no ponerlo por aquí a la vista de todo Internet!

Un saludo y me alegra ver (con envidia) que sigues operativo con el audio
Jose

Server listening with high level classes (asyncio), not socket.

A pre.di.c machine that plays movies sound or surround recordings would be great.

Hello I have prepared do_config_files.py: a loudspeaker config helper script.

You can have a human readable YAML file where to define your multiway loudspeaker configuration:

loudspeakers/example2ways+sub/example2ways+sub.yml

Then the script generates brutefir_config and speaker.yml to pre.di.c to work:

~$ do_config_files.py

 Usage:     do_config_files.py   loudspeakers/myLoudspeaker/  [ -w ] [ |less ]

            -w   Overwrites the existing 'speaker.yml' and 'brutefir_config',
                 otherwise '____.candidate' files will be provided.

 This generates low level files needed for pre.di.c to work, i.e.:

   'brutefir_config', 'speaker.yml'

 The following files are needed under the folder 'myLoudspeaker':

   - 'myLoudspeaker.yml':
     The complete full range, multiway, subwoofer loudspeaker definitions including
     pcm names, gain, delay and polarity, as well as target values for SLP and room EQ.

   - 'brutefir_settings.yml':
     General runtime options for Brutefir.

   -  and some '*.pcm' FIR files to be used for XO and DRC.

      (i) CONVENTIONS FOR NAMING *.pcm FIR FILES:

        XO:
          xo.someDescription.pcm

        DRC:
          drc.X_someDescription.pcm     Where X is the channel 'L' or 'R'

Some stuff 'ref_level_gain' and 'target_xxx_curve parameters' still need to be considered manually under speaker.yml.

Hope this helps.

I have renamed scripts/mpd.py to scripts/mpd_script.py because the 'import mpd' statement does not fails if the python-mpd package were not installed, i.e. the fail will be masked. So no warnings abuot missing package will appear for us to notice it.

Something that talks with the running server, instead of a standalone program that has great delays in loading and importing everything needed.

Target curves could be on frd format, and hence a separate frequencies file would be unneccesary.
All eq curves could be calculated on the fly for frequencies in the target file.
A program to interpolate targets on system frequency points from FRD curves could be handy.

Rewrite full DVB client as a server (state file: state_DVB.yml) and jack loop
creator.

A little mod will permit to comment out a whole line or something at the line tail in config/scripts

207             for line in [ x for x in open(bp.script_list_path) 
208                                 if not '#' in x.strip()[0] ]:

As it is already available none for DRC sets inside speaker.yml, it is useful to allow no filtering for some use cases, for example for a full range pasive loudspeaker.

DRC:
  filters: [f.drc.L, f.drc.R]
  sets:
    nearfield: [L_DynA42, R_DynA42]
    none: [none, none]
XO:
  filters: [f.fr.L, f.fr.R]
  sets:
    normal: [DynA42_left, DynA42_right]
    none: [none, none]

This is the code proposal to allow this under the function server_process.change_xovers :

            if XO_set in gc.speaker['XO']['sets']:
                coeffs = gc.speaker['XO']['sets'][XO_set]
                filters = gc.speaker['XO']['filters']
                # allows no filtering e.g for fullrange loudspeakers
                coeffsTmp = coeffs.copy() # a copy does not modify the original
                for index, item in enumerate(coeffsTmp):
                    if item == 'none':
                        coeffsTmp[index] = '-1'
                    else:
                        coeffsTmp[index] = '"' + item + '"'
                for i in range(len(filters)):
                    bf_cli( 'cfc "' + filters[i] + '" ' + coeffsTmp[i] )

The mpd->predic part works.

Hi @Rsantct : thanks for your efforts on the web page.

When loading predic this error arises, here reproduced manually:
predic@predic:~$ scripts/server_aux.py start
(server_aux) bad option

The result is that new functionalities of player communications do not work.

Also, on this line of loading script:
Popen( ['server_aux.py'] ) # if this fails check your paths
you could put the full path with the help of bin/basepaths.py

import basepaths as bp
---
path = bp.bin_folder

or something like that.

In what relates to a external switcher, any input on it could have an entry in inputs.yml, and a external command parameter could be added to the input entry for actual switching outside the jack realm.

There are wired paths to (now) predic folder, config files needs update YAML format, and a function in server_process.py is needed.

@rripio says:

Also, I think that we should discuss in advance a bit changes this big, that affects general distribution of code. That is, I would have rather have each player's specific code confined to it's own load script as much as possible.

That goes also for any auxiliary functions, and so leave the bin folder dedicated to the server, that is player agnostic.

Layers of functionality are to be as independent as possible from the server.

Originally posted by @rripio in #28 (comment)

in general, server response is messed up (by me), in absence of clients to test its validity, usefulness and completion.

selecting input "none" raise an exception in the server, related with gain routines being unable to retrieve the associated gain.

When startaudio.py completes, the status shows the last DRC_set in state.yml, but it is not running.

This mod inside startaudio.py fixes that behavoir:

def init_state_settings():
    """restore audio settings as stored in state.yaml
       and takes care of options to reset some of them
    """
    ...
    ...
    # restore DRC_set
    pd.client_socket( 'drc ' + str( gc.state['DRC_set'] ) )

    # XO_set will be adjusted when restoring inputs

I propose to include a little function update_asoundrc() inside startaudio.init_jack() to manage this issue. It can be found here.

Too much of them. See how to re-implement wait4() for this cases.

Hi @Rsantct , DVB.py script is broken by this merge, could you please review the script and update it?

Originally posted by @rripio in #27 (comment)

Gain correction for multi-input external modules (DVB, an analog input
switcher, etc.) as a server command.

Just a note: maybe wrap relevant code in a generic exception trap that restores state on any
command error.

When issue the command input none a warning appears:

Warnings:
	 Something went wrong when changing input state

It seems that it is missed an early return inside server_process.change_input():

    def change_input(input, state=state):

        state['input'] = input
        # if none disconnects all inputs
        if input == 'none':
            disconnect_inputs()
            return state                    <===== this one fix the problem

Calculate recording EQ on-the-fly and load to brutefir shared memory instead of generating brutefir graphic EQ settings.