[Sound-open-firmware] Introduction and questions
I write a column on open source and music on https://opensource.com/ Recently I was at Open Summit and learned of this project. I subsequently contacted some folks at Intel to learn more about how this project might affect users of open source media players. One of my contacts advised posting questions to this list. To avoid spamming the list any more than I already have, I'd like to confirm that a few such questions won't be a supreme annoyance to the list members.
Thanks in advance for any advice.
Hi Chris,
On Tue, 2018-09-11 at 14:12 -0700, chris hermansen wrote:
I write a column on open source and music on https://opensource.com/ Recently I was at Open Summit and learned of this project. I subsequently contacted some folks at Intel to learn more about how this project might affect users of open source media players. One of my contacts advised posting questions to this list. To avoid spamming the list any more than I already have, I'd like to confirm that a few such questions won't be a supreme annoyance to the list members.
Thanks in advance for any advice.
No annoyance will occur :) Feel free to ask. The reason I ask for the list is that the questions and answers will be all searchable.
Thanks
Liam
Liam and list,
Thanks for the kind reply, this is exciting!
On Wed, Sep 12, 2018 at 1:34 PM Liam Girdwood < liam.r.girdwood@linux.intel.com> wrote:
Hi Chris,
On Tue, 2018-09-11 at 14:12 -0700, chris hermansen wrote:
I write a column on open source and music on https://opensource.com/ Recently I was at Open Summit and learned of this project. I
subsequently
contacted some folks at Intel to learn more about how this project might affect users of open source media players. One of my contacts advised posting questions to this list. To avoid spamming the list any more
than I
already have, I'd like to confirm that a few such questions won't be a supreme annoyance to the list members.
Thanks in advance for any advice.
No annoyance will occur :) Feel free to ask. The reason I ask for the list is that the questions and answers will be all searchable.
Starting with a bit of background... there is a class of audio hardware and software user out there who uses a configuration like this:
1. laptop, desktop, server appliance... 2. running Linux 3. with a dedicated DAC used for music playback 4. connected to a high quality headphone amplifier or home stereo 5. with an open source music player client/server installed that probably accesses ALSA directly 6. with a bunch of music files, probably FLAC, maybe DSF, probably CD quality or higher 7. time on their hands to enjoy this
If the computer hardware itself is used for many and various things, then likely the above configuration will want to run most audio traffic through Pulse and out the built-in audio speakers / headphone, with only the high-quality music being sent to the dedicated DAC directly over ALSA to avoid software mixing, resampling etc. However if the computer hardware is more or less dedicated to music enjoyment, then it seems reasonable to hope that the built-in audio hardware could itself be dedicated to the primary music reproduction role, in a high-quality, low noise, low distortion, "bit-perfect" fashion.
So, looking at the sound-open-firmware initiative:
1. is this primarily oriented toward opening up the built-in audio processing chain? or should we expect add-on ADC and DAC hardware? 2. is it reasonable to expect really high-performance DAC implementations (low noise etc) coming out of this? 3. what kind of other applications might we expect to see? 1. could we expect to see the ability to build a digital crossover with a FIR filter to drive high, medium and low frequency power amplifiers and speakers? 2. what about other types of digital filtering to compensate for room acoustic issues? 3. might we see some type of audio compensation to "pull the music out of the head" for headphone users? 4. what about decoding DSF directly, not just PCM? 5. will people build proprietary things on top of it (for example, could MQA use it to build an MQA decoder)?
That's all I can think of right now; thanks in advance!
On Thu, 2018-09-13 at 16:09 -0700, chris hermansen wrote:
Liam and list,
+ Seppo for audio processing.
Thanks for the kind reply, this is exciting!
On Wed, Sep 12, 2018 at 1:34 PM Liam Girdwood <liam.r.girdwood@linux.intel.com
wrote: Hi Chris,
On Tue, 2018-09-11 at 14:12 -0700, chris hermansen wrote:
I write a column on open source and music on https://opensource.com/ Recently I was at Open Summit and learned of this project. I subsequently contacted some folks at Intel to learn more about how this project might affect users of open source media players. One of my contacts advised posting questions to this list. To avoid spamming the list any more than
I
already have, I'd like to confirm that a few such questions won't be a supreme annoyance to the list members.
Thanks in advance for any advice.
No annoyance will occur :) Feel free to ask. The reason I ask for the list is that the questions and answers will be all searchable.
Starting with a bit of background... there is a class of audio hardware and software user out there who uses a configuration like this: laptop, desktop, server appliance... running Linux with a dedicated DAC used for music playback connected to a high quality headphone amplifier or home stereo with an open source music player client/server installed that probably accesses ALSA directly with a bunch of music files, probably FLAC, maybe DSF, probably CD quality or higher time on their hands to enjoy this If the computer hardware itself is used for many and various things, then likely the above configuration will want to run most audio traffic through Pulse and out the built-in audio speakers / headphone, with only the high- quality music being sent to the dedicated DAC directly over ALSA to avoid software mixing, resampling etc. However if the computer hardware is more or less dedicated to music enjoyment, then it seems reasonable to hope that the built-in audio hardware could itself be dedicated to the primary music reproduction role, in a high-quality, low noise, low distortion, "bit-perfect" fashion.
also low power and low latency.
So, looking at the sound-open-firmware initiative: is this primarily oriented toward opening up the built-in audio processing chain?
Yes, SOF is infrastructure that allows processing components or pipelines to be constructed.
or should we expect add-on ADC and DAC hardware?
No, it's intended to work with all standard DAC and ADC hardware.
is it reasonable to expect really high-performance DAC implementations (low noise etc) coming out of this?
No this is a SW/FW project, but it is intended to use high quality DAC HW with SOF.
Liam
what kind of other applications might we expect to see? could we expect to see the ability to build a digital crossover with a FIR filter to drive high, medium and low frequency power amplifiers and speakers? what about other types of digital filtering to compensate for room acoustic issues? might we see some type of audio compensation to "pull the music out of the head" for headphone users? what about decoding DSF directly, not just PCM? will people build proprietary things on top of it (for example, could MQA use it to build an MQA decoder)? That's all I can think of right now; thanks in advance! -- Chris Hermansen · clhermansen "at" gmail "dot" com
C'est ma façon de parler.
Hi,
On 17.09.2018 21:53, Liam Girdwood wrote:
On Thu, 2018-09-13 at 16:09 -0700, chris hermansen wrote:
Liam and list,
- Seppo for audio processing.
Thanks for the kind reply, this is exciting!
On Wed, Sep 12, 2018 at 1:34 PM Liam Girdwood <liam.r.girdwood@linux.intel.com
wrote: Hi Chris,
On Tue, 2018-09-11 at 14:12 -0700, chris hermansen wrote:
I write a column on open source and music on https://opensource.com/ Recently I was at Open Summit and learned of this project. I subsequently contacted some folks at Intel to learn more about how this project might affect users of open source media players. One of my contacts advised posting questions to this list. To avoid spamming the list any more than
I
already have, I'd like to confirm that a few such questions won't be a supreme annoyance to the list members.
Thanks in advance for any advice.
No annoyance will occur :) Feel free to ask. The reason I ask for the list is that the questions and answers will be all searchable.
Starting with a bit of background... there is a class of audio hardware and software user out there who uses a configuration like this: laptop, desktop, server appliance... running Linux with a dedicated DAC used for music playback connected to a high quality headphone amplifier or home stereo with an open source music player client/server installed that probably accesses ALSA directly with a bunch of music files, probably FLAC, maybe DSF, probably CD quality or higher time on their hands to enjoy this If the computer hardware itself is used for many and various things, then likely the above configuration will want to run most audio traffic through Pulse and out the built-in audio speakers / headphone, with only the high- quality music being sent to the dedicated DAC directly over ALSA to avoid software mixing, resampling etc. However if the computer hardware is more or less dedicated to music enjoyment, then it seems reasonable to hope that the built-in audio hardware could itself be dedicated to the primary music reproduction role, in a high-quality, low noise, low distortion, "bit-perfect" fashion.
also low power and low latency.
So, looking at the sound-open-firmware initiative: is this primarily oriented toward opening up the built-in audio processing chain?
Yes, SOF is infrastructure that allows processing components or pipelines to be constructed.
or should we expect add-on ADC and DAC hardware?
No, it's intended to work with all standard DAC and ADC hardware.
is it reasonable to expect really high-performance DAC implementations (low noise etc) coming out of this?
No this is a SW/FW project, but it is intended to use high quality DAC HW with SOF.
Liam
what kind of other applications might we expect to see?
At the moment we are focusing to basics with audio processing components and enablers for development. They are test vehicles for developing IPC, driver, topology, and user space controls for customer's more complex applications. We are following customer input for our development of open source processing components examples.
could we expect to see the ability to build a digital crossover with a FIR filter to drive high, medium and low frequency power amplifiers and speakers?
Yes, though not completely with the current components set. This would possible to do as DIY with a fairly small effort. We welcome the community to develop new components and contribute and propose enhancements for existing components needed for their applications.
There would be for this particular application need for a topology controlled stream splitting component suitable for the DAC endpoints (multiple I2S, TDM, etc.). SOF currently has basic digital filters (FIR, IIR) implementations and we are at the moment working with kernel services for user space control for them. SOF is modular and the pipeline is defined in the user space so this would be mostly a new processing pipeline definition and suitable filters design task. There's also some filter design examples in SOFT (the SOF Tools) repository where design for this kind of need might be already be possible.
what about other types of digital filtering to compensate for room acoustic issues?
Yes, the existing filters are suitable for frequency response enhancement if the needed correction is known. Measuring and knowing the room is the beef / key challenge here.
This would be an interesting community project, I think.
might we see some type of audio compensation to "pull the music out of the head" for headphone users?
It's possible as relatively small DIY effort with the FIR filter and high level processing pipeline definition if the needed impulse responses (room, HRTF) and filters network to achieve this are known.
what about decoding DSF directly, not just PCM?
We do not have example for offloaded audio coding yet but generally it has been prepared for and it will be possible. SOF is all free and open so I'd assume the first examples will be codecs from similar origins (I'm not sure but possibly like Vorbis and FLAC).
will people build proprietary things on top of it (for example, could MQA use it to build an MQA decoder)?
Sure, all is possible ranging from development into own hobby use to licensing libraries to create commercial products thanks to the BSD license that applies SOF. Though with more complex applications like streaming there's need to consider pros and cons for partitioning things between platform CPU and DSP especially what IPC is needed between system parts what is the system gain for doing things on the DSP side.
Thanks, Seppo
That's all I can think of right now; thanks in advance!
Chris Hermansen · clhermansen "at" gmail "dot" com
C'est ma façon de parler.
Sound-open-firmware mailing list Sound-open-firmware@alsa-project.org http://mailman.alsa-project.org/mailman/listinfo/sound-open-firmware
Thanks very much, Liam and Seppo. I'm going to mull this over and see if I can pull an article out of this.
On Mon, Sep 17, 2018 at 11:54 AM Liam Girdwood < liam.r.girdwood@linux.intel.com> wrote:
On Thu, 2018-09-13 at 16:09 -0700, chris hermansen wrote:
Liam and list,
- Seppo for audio processing.
Thanks for the kind reply, this is exciting!
On Wed, Sep 12, 2018 at 1:34 PM Liam Girdwood <
liam.r.girdwood@linux.intel.com
wrote: Hi Chris,
On Tue, 2018-09-11 at 14:12 -0700, chris hermansen wrote:
I write a column on open source and music on https://opensource.com/ Recently I was at Open Summit and learned of this project. I
subsequently
contacted some folks at Intel to learn more about how this project
might
affect users of open source media players. One of my contacts
advised
posting questions to this list. To avoid spamming the list any more
than
I
already have, I'd like to confirm that a few such questions won't be
a
supreme annoyance to the list members.
Thanks in advance for any advice.
No annoyance will occur :) Feel free to ask. The reason I ask for the
list
is that the questions and answers will be all searchable.
Starting with a bit of background... there is a class of audio hardware
and
software user out there who uses a configuration like this: laptop, desktop, server appliance... running Linux with a dedicated DAC used for music playback connected to a high quality headphone amplifier or home stereo with an open source music player client/server installed that probably accesses ALSA directly with a bunch of music files, probably FLAC, maybe DSF, probably CD
quality or
higher time on their hands to enjoy this If the computer hardware itself is used for many and various things, then likely the above configuration will want to run most audio traffic
through
Pulse and out the built-in audio speakers / headphone, with only the
high-
quality music being sent to the dedicated DAC directly over ALSA to avoid software mixing, resampling etc. However if the computer hardware is
more or
less dedicated to music enjoyment, then it seems reasonable to hope that
the
built-in audio hardware could itself be dedicated to the primary music reproduction role, in a high-quality, low noise, low distortion,
"bit-perfect"
fashion.
also low power and low latency.
So, looking at the sound-open-firmware initiative: is this primarily oriented toward opening up the built-in audio
processing
chain?
Yes, SOF is infrastructure that allows processing components or pipelines to be constructed.
or should we expect add-on ADC and DAC hardware?
No, it's intended to work with all standard DAC and ADC hardware.
is it reasonable to expect really high-performance DAC implementations
(low
noise etc) coming out of this?
No this is a SW/FW project, but it is intended to use high quality DAC HW with SOF.
Liam
what kind of other applications might we expect to see? could we expect to see the ability to build a digital crossover with a
FIR
filter to drive high, medium and low frequency power amplifiers and
speakers?
what about other types of digital filtering to compensate for room
acoustic
issues? might we see some type of audio compensation to "pull the music out of
the
head" for headphone users? what about decoding DSF directly, not just PCM? will people build proprietary things on top of it (for example, could
MQA use
it to build an MQA decoder)? That's all I can think of right now; thanks in advance! -- Chris Hermansen · clhermansen "at" gmail "dot" com
C'est ma façon de parler.
participants (3)
-
chris hermansen -
Liam Girdwood -
Seppo Ingalsuo