Achieving maximum audio quality?

[esologic]

At the outset I must say that this project feels like magic. Even with my status as a complete novice to the world of networked audio, I was able to get point-to-point audio transmission over my network between a pair of raspberry pi's in an afternoon. My setup is as follows:

Nakamichi 600 Tape Deck -> Behringer UMC404HD -> Raspberry Pi -> LAN -> Raspberry Pi -> Audio Receiver -> Speakers.

This works just fine for now and has the advertised low latency. What I'm having trouble with is increasing the quality of the transmission.

My roc-send incantation:

./roc-send -vv -s rtp+rs8m://192.168.1.118:10001 -r rs8m://192.168.1.118:10002 -c rtcp://192.168.1.118:10003 --input alsa://stereo_downmix --rate=192000 --max-frame-size=20000

The corresponding roc-recv:

./roc-recv -vv -s rtp+rs8m://localhost:10001 -r rs8m://localhost:10002 -c rtcp://localhost:10003 -o alsa://hw:0,0

You'll wonder about the stereo_downmix thing -- this is because roc-send didn't like the 4 channels emitted from UMC404HD:

pi@audio-pi-server:~/roc/roc-toolkit/bin/arm-pc-linux-gnueabihf $ ./roc-send -vv -s rtp+rs8m://192.168.1.118:10001 -r rs8m://192.168.1.118:10002 -c rtcp://192.168.1.118:10003 --input alsa://hw:2,0 --rate=192000 --max-frame-size=20000
11:42:19.095 [743] [dbg] roc_sndio: [sndfile_backend.cpp:20] sndfile backend: initializing
11:42:19.197 [743] [dbg] roc_sndio: [backend_map.cpp:22] backend map: initializing: n_backends=4 n_drivers=109
11:42:19.197 [743] [dbg] roc_core: [slab_pool_impl.cpp:63] slab pool (packet_pool): initializing: slot_size=496 prealloc_size=8(0 slots) min_slab=8(1 slots) max_slab=0(0 slots)
11:42:19.197 [743] [dbg] roc_core: [slab_pool_impl.cpp:63] slab pool (packet_buffer_pool): initializing: slot_size=46184 prealloc_size=8(0 slots) min_slab=8(1 slots) max_slab=0(0 slots)
11:42:19.197 [743] [dbg] roc_core: [slab_pool_impl.cpp:63] slab pool (frame_buffer_pool): initializing: slot_size=20040 prealloc_size=8(0 slots) min_slab=8(1 slots) max_slab=0(0 slots)
11:42:19.197 [743] [dbg] roc_core: [slab_pool_impl.cpp:63] slab pool (encoding_pool): initializing: slot_size=240 prealloc_size=3840(16 slots) min_slab=3840(16 slots) max_slab=0(0 slots)
11:42:19.202 [744] [dbg] roc_netio: [network_loop.cpp:277] network loop: starting event loop
11:42:19.202 [745] [dbg] roc_ctl: [control_task_queue.cpp:95] control task queue: starting event loop
11:42:19.202 [743] [dbg] roc_node: [context.cpp:26] context: initializing
11:42:19.202 [743] [inf] roc_sndio: [sox_source.cpp:79] sox source: opening: driver=alsa path=hw:2,0
11:42:19.661 [743] [dbg] roc_sndio: [sox_backend.cpp:157] sox: formats.c: can't set 2 channels; using 4
11:42:19.661 [743] [err] roc_sndio: [sox_source.cpp:385] sox source: can't open input file or device with the requested channel count: required_by_input=4 requested_by_user=2
11:42:19.661 [743] [dbg] roc_sndio: [sox_backend.cpp:284] sox backend: open failed: driver=alsa path=hw:2,0
11:42:19.661 [743] [inf] roc_sndio: [sox_source.cpp:414] sox source: closing input
11:42:19.662 [743] [err] roc_sndio: [backend_dispatcher.cpp:233] backend dispatcher: failed to open source: type=device driver=alsa path=hw:2,0
11:42:19.662 [743] [err] roc_send: [main.cpp:315] can't open input file or device: uri=alsa://hw:2,0 format=(null)
11:42:19.662 [743] [dbg] roc_node: [context.cpp:30] context: deinitializing
11:42:19.662 [745] [dbg] roc_ctl: [control_task_queue.cpp:105] control task queue: finishing event loop
11:42:19.663 [744] [dbg] roc_netio: [network_loop.cpp:284] network loop: finishing event loop

So I created a ~/.asoundrc with:

pi@audio-pi-server:~/roc/roc-toolkit/bin/arm-pc-linux-gnueabihf $ cat ~/.asoundrc
pcm.stereo_downmix {
    type plug
    slave.pcm "hw:2,0"
    ttable.0.0 1
    ttable.1.1 1
}

Okay so my question -- If I don't really care about latency, which command line settings should I tweak to max out quality?

Reading the docs, I see a lot of knobs to turn but don't really understand what effect they'll have. For example:

--resampler-profile=ENUM
Resampler profile (possible values=”low”, “medium”, “high” default=`medium’)

I can't really tell which setting leads to more data here. Here's some pactl list sources output for UMC404HD for context.

pi@audio-pi-server:~/roc/roc-toolkit/bin/arm-pc-linux-gnueabihf $ pactl list sources
Source #1
	State: SUSPENDED
	Name: alsa_output.usb-BEHRINGER_UMC404HD_192k-00.analog-surround-40.monitor
	Description: Monitor of UMC404HD 192k Analog Surround 4.0
	Driver: module-alsa-card.c
	Sample Specification: s16le 4ch 44100Hz
	Channel Map: front-left,front-right,rear-left,rear-right
	Owner Module: 8
	Mute: no
	Volume: front-left: 65536 / 100% / 0.00 dB,   front-right: 65536 / 100% / 0.00 dB,   rear-left: 65536 / 100% / 0.00 dB,   rear-right: 65536 / 100% / 0.00 dB
	        balance 0.00
	Base Volume: 65536 / 100% / 0.00 dB
	Monitor of Sink: alsa_output.usb-BEHRINGER_UMC404HD_192k-00.analog-surround-40
	Latency: 0 usec, configured 0 usec
	Flags: DECIBEL_VOLUME LATENCY 
	Properties:
		device.description = "Monitor of UMC404HD 192k Analog Surround 4.0"
		device.class = "monitor"
		alsa.card = "2"
		alsa.card_name = "UMC404HD 192k"
		alsa.long_card_name = "BEHRINGER UMC404HD 192k at usb-3f980000.usb-1.2, high speed"
		alsa.driver_name = "snd_usb_audio"
		device.bus_path = "platform-3f980000.usb-usb-0:1.2:1.0"
		sysfs.path = "/devices/platform/soc/3f980000.usb/usb1/1-1/1-1.2/1-1.2:1.0/sound/card2"
		udev.id = "usb-BEHRINGER_UMC404HD_192k-00"
		device.bus = "usb"
		device.vendor.id = "1397"
		device.vendor.name = "BEHRINGER International GmbH"
		device.product.id = "0509"
		device.product.name = "UMC404HD 192k"
		device.serial = "BEHRINGER_UMC404HD_192k"
		device.string = "2"
		module-udev-detect.discovered = "1"
		device.icon_name = "audio-card-usb"
	Formats:
		pcm

Source #2
	State: SUSPENDED
	Name: alsa_input.usb-BEHRINGER_UMC404HD_192k-00.multichannel-input
	Description: UMC404HD 192k Multichannel
	Driver: module-alsa-card.c
	Sample Specification: s32le 4ch 44100Hz
	Channel Map: front-left,front-right,rear-left,rear-right
	Owner Module: 8
	Mute: no
	Volume: front-left: 65536 / 100% / 0.00 dB,   front-right: 65536 / 100% / 0.00 dB,   rear-left: 65536 / 100% / 0.00 dB,   rear-right: 65536 / 100% / 0.00 dB
	        balance 0.00
	Base Volume: 65536 / 100% / 0.00 dB
	Monitor of Sink: n/a
	Latency: 0 usec, configured 0 usec
	Flags: HARDWARE DECIBEL_VOLUME LATENCY 
	Properties:
		alsa.resolution_bits = "32"
		device.api = "alsa"
		device.class = "sound"
		alsa.class = "generic"
		alsa.subclass = "generic-mix"
		alsa.name = "USB Audio"
		alsa.id = "USB Audio"
		alsa.subdevice = "0"
		alsa.subdevice_name = "subdevice #0"
		alsa.device = "0"
		alsa.card = "2"
		alsa.card_name = "UMC404HD 192k"
		alsa.long_card_name = "BEHRINGER UMC404HD 192k at usb-3f980000.usb-1.2, high speed"
		alsa.driver_name = "snd_usb_audio"
		device.bus_path = "platform-3f980000.usb-usb-0:1.2:1.0"
		sysfs.path = "/devices/platform/soc/3f980000.usb/usb1/1-1/1-1.2/1-1.2:1.0/sound/card2"
		udev.id = "usb-BEHRINGER_UMC404HD_192k-00"
		device.bus = "usb"
		device.vendor.id = "1397"
		device.vendor.name = "BEHRINGER International GmbH"
		device.product.id = "0509"
		device.product.name = "UMC404HD 192k"
		device.serial = "BEHRINGER_UMC404HD_192k"
		device.string = "hw:2"
		device.buffering.buffer_size = "42304"
		device.buffering.fragment_size = "10576"
		device.access_mode = "mmap"
		device.profile.name = "multichannel-input"
		device.profile.description = "Multichannel"
		device.description = "UMC404HD 192k Multichannel"
		module-udev-detect.discovered = "1"
		device.icon_name = "audio-card-usb"
	Formats:
		pcm

Thank you so much!

[baranovmv]

Hi @esologic, I'm glad you found roc useful.

First of all, in current master branch there is not many options you can tune related with quality. Resampler's profile would be the one of them, try setting it to high roc-recv ....... --resampler-profile=high and check cpu usage on the receiver side, if it is ok for you, and no glitches occurs, then leave it. Otherwise medium profile is actually also decent. Also, I'd recommend to fix latency on some sensibly big fixed value, which could very much depend on network you are using. If it is wireless, I'd suggest go several hundreds of ms: --target-latency=300ms. In case of wired connection, safe guess would be 50 ms. But these values are just uninformed suggestions, they might not fit your circumstances, e.g. if WiFi channel is very much occupied by your neighbors.

Then, the audio samples transfered by roc through your LAN in internal fixed format of 16 bit stereo @ 44.1 kHz (@gavv, pls correct me). However, recently some changes were made in that regard, which are presented nowadays in develop branch. Sample bit-width, available channel number and there mapping, and sample rate of the payload format could be set for roc-send and roc-recv simultaneously in command line, which leads to even more available tweaks and knobs, sorry for that. Unfortunately, it will require for you to build roc yourself (may be done on raspberry, if it is one of the recent versions).

I haven't occasion to try this option myself yet, so i if you are interested in building develop, please let me know and I'll check how/if it works.