Helloguys,
IamdevelopinganAPOforacustomaudioprocessingeffect.Likemany
othersIhavebeganwithsysvad.
AfterovercomingsomebasicissuesIhavesuccessfullybuiltallmodules
usingVS15.
Inowhavethefollowingmodulesbuilt:
KeywordDetectorContosoAdapter.dll
PhoneAudioSample.sys
...
[wdmaudiodev] some issues with sysvad sample - help appreciated
↧
↧
[wdmaudiodev] Re: some issues with sysvad sample - help appreciated
Theprojectthatbuilttabletaudiosample.sysshouldalsohave“built”(really,
stamped)tabletaudiosample.inf.
From:GilShefer
Sent:Saturday,April1,20172:42PM
To:wdmaudiodev@xxxxxxxxxxxxx
Subject:[wdmaudiodev]someissueswithsysvadsample-helpappreciated
Helloguys,
...
↧
[wdmaudiodev] Re: Where to report issues with the Windows 10 UAC2 audio driver?
Let’ssee…
Yes,thedevshaveaskedifthisreprosonthelatestInsiderbuild
(currentlythelatestInsiderbuildis16299)
Yes,it’sexactlythesamebehaviorwithInsiderbuild16299.ShouldIprobably
openasupportincidenttogetincontactwiththedevelopers?
...
↧
[wdmaudiodev] Re: Where to report issues with the Windows 10 UAC2 audio driver?
Thanks,Ihaveconveyedthatinformationtothedevelopers.
Youcouldcertainlyopenasupportcaseifyouwantedtoexpeditethings.If
youdo,pleasetellthesupportengineerthisisabout“OSbug13834139.”
From:FranzDetro
Sent:Tuesday,October10,201711:35PM
To:wdmaudiodev@xxxxxxxxxxxxx
Subject:[wdmaudiodev]Re:WheretoreportissueswiththeWindows10UAC2
audiodriver?
...
↧
[wdmaudiodev] Re: Windows Real Time : suggestion to Microsoft Kernel Developers
Hello,
OneyearafterIcomebacktothissuggestion,toturnWindowsinaRTmode
(enoughRTforRTaudioImean),
afterhavingthoughttodifferentpossibilitieslikethreadRT
characteristics,CoreAllocationetc
.
...
↧
↧
[wdmaudiodev] Re: Windows Real Time : suggestion to Microsoft Kernel Developers
CanyousubmitthisasasuggestionviaFeedbackHub?
________________________________
From:wdmaudiodev-bounce@xxxxxxxxxxxxxwdmaudiodev-bounce@xxxxxxxxxxxxxon
behalfofVincentBurel(VB-Audio)vincent.burel@xxxxxxxxxxxx
Sent:Thursday,October12,201712:49:03AM
To:wdmaudiodev@xxxxxxxxxxxxx
Subject:[wdmaudiodev]Re:WindowsRealTime:suggestiontoMicrosoftKernel
Developers
...
↧
[wdmaudiodev] Re: Windows Real Time : suggestion to Microsoft Kernel Developers
Idontknowthis
couldnotitbesimplertoforwardthismailtotherightspeopleinstead
?
Thanksbyadvance
Regards
VincentBurel
...
↧
[wdmaudiodev] Re: Windows Real Time : suggestion to Microsoft Kernel Developers
No,sorry
________________________________
From:wdmaudiodev-bounce@xxxxxxxxxxxxxwdmaudiodev-bounce@xxxxxxxxxxxxxon
behalfofVincentBurel(VB-Audio)vincent.burel@xxxxxxxxxxxx
Sent:Saturday,October14,20172:30:16AM
To:wdmaudiodev@xxxxxxxxxxxxx
Subject:[wdmaudiodev]Re:WindowsRealTime:suggestiontoMicrosoftKernel
Developers
...
↧
[wdmaudiodev] Windows Clock Compensation
Weallknowthatcrystalsarenotperfect. Evenifbothendsofan
audioconversationagreetoanominal48kHzsamplingrate,witha100ppm
crystal,theactualratemightbebetween47,995and48,005samplesper
second. Andiftheinputandtheoutputareondifferentboards,they
mightbeatoppositeendsofthatrange. Andyet,wedonothear
complaintsfromusersofcheapcommodityhardwareaboutperiodic
glitchesintheirvideoconferences.
Whyisthat? IstheAudioEnginecompensatingforthis? Isit
monitoringaveragebyte-per-secondratesoverlongperiodsanddoing
...
↧
↧
[wdmaudiodev] Re: Windows Clock Compensation
Severalthings:
1.Ofteninputsandoutputsarefromandtothesameaudioperipheral,which
usesthesameclockforrenderandcapture.Soalthoughtheclockmaybeoff
(e.g.,47.995kHzinsteadoftrue48kHz),boththeinputclockandtheoutput
clockareoffinthesamedirectionbythesameamount.
2.Incaseslikevideoconferencingwheretheinputandoutputareon
totallydifferentdevice,skewcompensationisdoneinsoftware.Forexample,
...
↧
[wdmaudiodev] Re: Windows Clock Compensation
Firmwareengineerhere.
One possible solution we implemented for usb-audio using synchronous
mode was to adjust the reference clocks on the micro-controller in
realtime by a very small amount to ensure that there are no
underflow/overflow conditions that manifest themselves as pops/clicks(we
did have pops and clicks otherwise)
...
↧
[wdmaudiodev] Re: Windows Clock Compensation
UACinheritsfromUSBbasespec. Specifically,forthistopic,ISOtransfer
types::Async,Sync,Adaptive. ItistheresponsibilityforUACdeviceto
maintainbuffermanagement(onemethodoranother)::
-ISOAsync-explicitvsimplicitfeedbackmechanism---audiobuffersare
maintainedviathedevice explicitlytellingthehostthatitisonesample
ahead(-a),onesamplebehind(+1),orinlockstep(0)viaseparateISO
feedbackendpoint,orimpliedbythebehavioroftherecordingstream(dynamic
slightadjustmentsasnotedforexplicit)
...
↧
[wdmaudiodev] Re: Windows Clock Compensation
AkshaykeertiSharmawrote:
Onepossiblesolutionweimplementedforusb-audiousingsynchronous
modewastoadjustthereferenceclocksonthemicro-controllerin
realtimebyaverysmallamounttoensurethatthereareno
underflow/overflowconditionsthatmanifestthemselvesas
pops/clicks(wedidhavepopsandclicksotherwise)
Howwouldyouknow? BymonitoringthelevelsinyouronboardFIFOs?
Thisisanextremelytrickyproblem,becauseeventhoughaudioseemsto
...
↧
↧
[wdmaudiodev] Re: Windows Clock Compensation
Idon'tknowhowothersdoit,butthisclocksynchronizationmethod(ISO
Sync)isveryeasytodoinhardware
yousimplyinfertheUSBhost'sclockbyaccuratelytimingSOFSOFinhardware
(notviaMCUISR)anddynamically(butslowly)adjustyourlocalmasteraudio
clocktomanageaudiobuffers...youcanachieveaveryaccuratewaterlevel
using thismethodof synchronization
ISOAsyncisacompletelydifferentsynchronizationmethod,buthasthe
advantageoflowjitteraudiomasterclock(since the synchronizationmethod
aboveisremovedand feedbackissolvedotherways)
It'sinnowway extremelytrickyproblem...but,intheearlydaysofUACfor
...
↧
[wdmaudiodev] Re: Windows Clock Compensation
Right,thisishowyouslaveyourbuffertransferstotheeventualADCorDAC
whiledealingwithjitter.
Tim’squestioniswhathappensifyouhaveapipelinewithanADCononeend,
andaDAContheother,andthereisskewbetweenthem.
________________________________
From:wdmaudiodev-bounce@xxxxxxxxxxxxxwdmaudiodev-bounce@xxxxxxxxxxxxxon
behalfofTroyGentrydmarc-noreply@xxxxxxxxxxxxx
Sent:Wednesday,October18,20173:38:39PM
...
↧
[wdmaudiodev] Re: Windows Clock Compensation
programmatically,I'ddiscoverthebDelaylatencyparameterfromtheADCand
fromtheDAC(assumingnotsamephysicaldevice)andusethosenumbersfor
worstcaseanalysis
typically,USBdeviceis goingtoreportanywherefrom1-3msecforbDelay....
PLUSthere'ssimilarbDelayforUSBhostsystem. Ithinksomesoftwarelike
ProLogicwilltellyouroundtripdelayforsame device,notsureaboutround
tripdelaytimingfordifferentdevices...
so,yeahabovemypaygradehere,sorryguys
From:MatthewvanEerdedmarc-noreply@xxxxxxxxxxxxx
...
↧
[wdmaudiodev] Re: Windows Clock Compensation
Yes,wearemonitoringthemicrocontrollerqueueslivetoensurethatthereis
noproblem.IfIremembercorrectlytheusbaudio1.0specweareusing
transmits48samples/ms(48khz)inourimplementation.Makingaverysmallchange
toourI2Sreferenceclock(thisiswhatdrivestheaudioouttoaspeaker
essentially)controlstheFIFOqueue.Thetrimsettingonthemicroisaccurate
enoughtoensurepreciseenoughchangestotheclocktosometimesevenmatch
theinputdatastream.
Anotherwayistosynctotheusbframestartofframewhichresultsingetting
thesameinformation.Youwouldsynctothestartframeandup/downtheclock
...
↧
↧
[wdmaudiodev] Re: Windows Clock Compensation
doingitthisway,isasTimsaysanextremelytrickyproblem-agreeddoing
itthisway,willlikelyproducenoticeableaudiblesynchronizationclocking
artifactsintheactualaudiostreamitself(dependingontheactualclock
mismatchbetweentheUSBHostandtheUSBdevice)
ifthisisallonehasfortheiraudiobuffermanagementmechanism,then
switchingtoASYNCsynchronizationmethodwouldbehighlyrecommended.
From:AkshaykeertiSharmaasharma@xxxxxxxxxxxxxxxxxxxxx
To:wdmaudiodev@xxxxxxxxxxxxx
Sent:Wednesday,October18,20174:51PM
...
↧
[wdmaudiodev] Re: Windows Clock Compensation
Ifyouhavetheproperhardware(i.e.afractionalPLL)theadjustmentsyoucan
maketothelocalclockareextremelyprecise(typically0.01ppm)Ifyouthen
alsoimplementalow-passfilterbehaviorontheclockadjustmentalgorithmand
deployveryextensiveaveraging,youcanachieveaudioclockqualitylevels
thatarebeyondwhatcurrentADCsandDACsrequire.
Goingwiththeasynchronousmodeactuallyshifts(notsolves)theproblemto
theHost.IftheHosthassufficientlysophisticatedasynchronoussamplingrate
algorithmsavailable,thenthatmaybeagoodsolution.Butdonotassumethat
allqualityconcernsevaporatewhenyouswitchtoasyncmode.
...
↧
[wdmaudiodev] Re: Windows Clock Compensation
inthereliesthetruth,the opertativephrasehereisIfyouhavetheproper
hardware...thepreviousposterisattemptingtosolvethisageoldproblem
mostlyinsoftware...also,aSOCthatnativelyhasanIntegerPLL(alongwith
afewothergoodies)canactually(hasactually)solvethisproblem,andfor
bothclocktrees,andfromasingleexternalcrystal...soonereallycanpull
arabbitoutofahat;)
Ihaven'tranintomanyearsthatcandothePepsichallengewithASYNCvs
SYNC(properlydoneinhardware,asGeertlaysoutbelow)andactuallytellthe
difference,butIhavehadtheprivilegeofmeetingonesuchsole(ataworld
renownedmusicstudio)thatcouldhearthedifference,itwastrulyanamazing
...
↧
More Pages to Explore .....