Wednesday, 16 November 2011< ^ >
gmaxwell has set the subject to: Codec WG meeting IETF-81 | | | Say MIC: to have your comments spoken in the room
Room Configuration

[01:43:24] rillian joins the room
[01:45:53] irc- joins the room
[01:45:53] <irc-> * joined: gmaxwell
[01:46:46] gmaxwell joins the room
[01:47:30] gmaxwell has set the subject to: Codec WG meeting IETF-82 | | Streaming? | Say MIC: to have your comments spoken in the room
[01:56:29] <irc-> * joined: rillian__
[02:07:22] irc- leaves the room
[04:33:40] Erik Norvell joins the room
[04:35:27] Ralph Giles joins the room
[04:36:12] tsuichi joins the room
[04:37:32] bemasc joins the room
[04:39:19] gmaxwell has set the subject to: Codec WG meeting IETF-82 | | Streaming rtsp:// Meetecho | Say MIC: to have your comments spoken in the room
[04:40:22] MeetechoAudioChannel joins the room
[04:42:14] Simon Romano joins the room
[04:42:29] irc- joins the room
[04:42:37] Ben Schwartz joins the room
[04:45:55] <bemasc> anyone know if the meetecho audio is lower latency than the .m3u mp3 stream?
[04:46:12] Ralph Giles leaves the room
[04:46:27] <Erik Norvell> I'm listening to both right now, and it appears the meetecho stream has a few seconds higher latency
[04:46:36] <bemasc> ok, thanks
[04:47:07] <Erik Norvell> no sorry, the other way around!
[04:47:21] <Simon Romano> Right!
[04:47:36] <Simon Romano> The Meetecho stream should be almost real-time.
[04:47:50] <Simon Romano> Slide 1: CODEC WG
[04:48:30] <Simon Romano> Meetecho is @
[04:48:53] <Simon Romano> RTSP audio is @ rtsp://
[04:50:28] Jonathan Rosenberg joins the room
[04:50:49] <bemasc> sigh. One day we will not have to choose between high latency and awful barely intelligible quality.
[04:52:22] <irc-> * joined: derf
[04:52:50] katwalsh joins the room
[04:53:21] <Jonathan Rosenberg> Hello all
[04:54:14] <irc-> * joined: mindspillage
[04:56:18] Ben Schwartz leaves the room
[04:56:35] <Simon Romano> Hi Jonathan...
[04:57:27] <Simon Romano> @bemasc: do you think the audio quality is really "awful barely intelligible"?
[04:57:37] Jeff Huang joins the room
[04:58:11] <Simon Romano> Presentation stopped
[04:58:12] <bemasc> Simon Romano: I think it will be fine for the purpose of the meeting.
[04:58:23] Lorenzo Miniero joins the room
[04:58:27] <Simon Romano> Slide 1: CODEC WG
[04:58:41] Christian Hoene joins the room
[04:59:05] <bemasc> I do think it's substantially less intelligible than the (>WB) MP3, but I still prefer it if it has even a little lower latency.
[05:01:56] <Christian Hoene> any audio yet?
[05:01:58] burn joins the room
[05:02:09] <Lorenzo Miniero> I'm listening
[05:02:11] <Simon Romano> They have not yet started...
[05:02:18] <Lorenzo Miniero> but could hear Cullen very far before
[05:02:29] <Lorenzo Miniero> don't know if he was far from the mic though
[05:02:29] <burn> hi group. I will be your jabber microphone relay today.
[05:02:45] <Lorenzo Miniero> ok now :)
[05:02:51] <Christian Hoene> yes :-)
[05:02:53] Pen joins the room
[05:02:55] Matthew Kaufman joins the room
[05:03:38] <Simon Romano> A slow start...
[05:03:58] victor.pascual.avila joins the room
[05:04:20] <Simon Romano> Slide 2: Note Well
[05:04:22] <Simon Romano> Slide 3: Agenda
[05:04:48] rillian leaves the room
[05:04:51] <Simon Romano> Slide 4: Liaisons
[05:05:18] Ralph Giles joins the room
[05:05:52] <Simon Romano> Presentation stopped
[05:05:58] <burn> any remote participants having trouble hearing?
[05:06:09] <Jonathan Rosenberg> is audio faint for others?
[05:06:22] bemasc hears the RTSP stream loud, if not clear
[05:06:30] <Lorenzo Miniero> audio is fine for me
[05:06:34] <Christian Hoene> I am using Java. It is ok
[05:06:37] <Lorenzo Miniero> I'm using RTSP by the way
[05:06:47] <Lorenzo Miniero> rtsp://
[05:07:30] <Simon Romano> Slide 1: draft-ietf-codec-results-00
[05:08:28] <Simon Romano> Slide 2: Diff to draft-valin-codec-results-00
[05:09:23] <Simon Romano> Slide 4: Narrowband Coding Mandarin Speech
[05:10:02] <Simon Romano> Slide 5: Wideband and fullband codecs Mandarin sp
[05:10:11] <Simon Romano> eech
[05:10:37] pcgod joins the room
[05:10:48] <Simon Romano> Slide 6: 100,0
[05:11:04] <Simon Romano> Narrowband transcoding
[05:11:10] <Simon Romano> English speech
[05:12:00] <Simon Romano> Slide 7: Wideband transcoding English speech
[05:12:25] <Simon Romano> Slide 8: Results of Universität Tübingen
[05:13:08] <Simon Romano> Slide 9: Reference Items
[05:13:36] <Simon Romano> Slide 10: Results: Codecs
[05:14:10] <Simon Romano> Slide 11: Codec and Item
[05:14:34] <Simon Romano> Slide 12: Summary of both tests
[05:15:38] <burn> Jean-Marc at mic
[05:16:14] <burn> ?? at mic (missed the name)
[05:16:28] <Erik Norvell> Paul Coverdale
[05:16:36] <burn> Erik, thx
[05:17:04] <Simon Romano> Slide 13: Next steps: Codec characterization
[05:18:29] RjS joins the room
[05:18:30] <Simon Romano> Slide 14: • Please send new listening-test resul
[05:18:37] <Christian Hoene> ok
[05:18:42] <Simon Romano> Presentation stopped
[05:18:53] <Simon Romano> Slide 1: draft-ietf-codec-opus-10 Comments and ch
[05:19:19] <Simon Romano> Slide 2: C Code (SG16)
[05:19:33] Jonathan Lennox joins the room
[05:19:39] <Simon Romano> Slide 1: draft-ietf-codec-opus-10 Comments and ch
[05:19:41] <Simon Romano> Slide 2: C Code (SG16)
[05:20:42] <Simon Romano> Slide 3: Portability (SG16)
[05:22:01] <Simon Romano> Slide 4: Time Stretching/Shortening (SG 16)
[05:22:28] Matt Kotvis joins the room
[05:22:38] <Simon Romano> Slide 5: Draft Issues (SG16, last meeting)
[05:23:09] <Simon Romano> Slide 6: Mode Switching (Anisse Taleb, old)
[05:23:57] <Simon Romano> Slide 7: Test Vectors (SG16, last meeting)
[05:24:21] <Simon Romano> I do not have this slide, sorry :-(
[05:24:33] <Simon Romano> They must have updated the deck right before the meeting.
[05:24:59] <Matthew Kaufman> I'm very sorry - the slides just got changed
[05:25:08] <gmaxwell> (It's worth mentioning that one of the test vectors switches between all pairs of full-band modes, with a synthetic buzz signal that was selected to make any switching glitches really obvious)
[05:25:22] <Simon Romano> Slide 8: Compliance (Erik Norvell)
[05:25:24] <Matthew Kaufman> We will upload the current ones as soon as I can get my computer back
[05:25:31] rillian joins the room
[05:25:36] <Simon Romano> :-)
[05:25:37] <burn> gmaxell, would you like that relayed at the mic? (or are you in the room)
[05:25:50] <Simon Romano> no problem...was just a minor modification, as far as I saw
[05:25:51] <gmaxwell> no. I'm not. I guess we've moved on.
[05:26:28] <irc-> * joined: xiphmont
[05:26:32] <Erik Norvell> MIC: Informative encoder/decoder does not render the entire standard informative. (As response to the bullet). The bitstream is still normative.
[05:26:34] <burn> gmaxwell: okay. just prefix with mic: and I'll take straight to the mic.
[05:26:55] <burn> erik: i'm at the mic waiting
[05:27:53] <Erik Norvell> thanks
[05:27:53] victor.pascual.avila leaves the room
[05:28:18] <Matthew Kaufman> Erik - if you want to say more here, let us know
[05:28:37] victor.pascual.avila joins the room
[05:28:37] <burn> matthew: took the words right out of my fingers :)
[05:29:23] Jeff Huang leaves the room
[05:30:57] <burn> Koen Vos at mic
[05:31:06] <burn> Previously at mic was Stephan Wenger
[05:31:07] victor.pascual.avila leaves the room
[05:31:56] Monty Montgomery joins the room
[05:31:58] <burn> Stephan back at mic
[05:32:13] victor.pascual.avila joins the room
[05:32:15] <bemasc> sensitivity can be relaxed a lot more easily than it can be tightened, IMHO
[05:32:33] <burn> bemasc: sorry, don't recognize your nick. who are you?
[05:32:42] bemasc is Ben Schwartz
[05:32:47] <burn> oh, duh
[05:33:09] victor.pascual.avila leaves the room
[05:33:12] <burn> if you want relayed at mic, let me know. otherwise i will assume just a local comment
[05:33:19] <Christian Hoene> A Comment to the MIC: Under the light of legal IPR requirements, I would suggest to make the tests less strict - maybe adding a perceptual PEAQ comparision.
[05:33:25] <Erik Norvell> MIC: If the test disqualifies imperceptible differences the constraint is clearly too tight.
[05:33:40] <burn> Koen Vos again at mic
[05:33:57] <burn> I am in line next for Christian and Erik (after Stephan who just jumped up)
[05:34:18] <gmaxwell> I don't believe anyone in the world has the perceptual prowess to pass all imperceptible differences without also allowing bad sounding differences.
[05:34:30] victor.pascual.avila joins the room
[05:34:55] victor.pascual.avila leaves the room
[05:34:59] sun joins the room
[05:35:27] <bemasc> mic: In the process of DRAFT->RFC, can we not relax the requirement if it proves too tight in practice?
[05:35:31] victor.pascual.avila joins the room
[05:35:55] victor.pascual.avila leaves the room
[05:36:00] <burn> stephan is next, then i will relay for you, Ben
[05:36:16] <bemasc> thanks
[05:36:42] <bemasc> burn: I guess I mean PROPOSED->STANDARD
[05:36:54] <burn> bemasc: okay
[05:38:01] victor.pascual.avila joins the room
[05:38:53] <Christian Hoene> MIC: Have one test vector for both fixed and float would make sense.
[05:39:05] <burn> ack Christian
[05:39:31] <Erik Norvell> MIC: The compliance test does not need to be a quality test. For certain platforms/optimization, even reducing quality should be allowed. Note that this is already the case for the encoder - it may change the quality in either way.
[05:39:38] <burn> ack Erik
[05:39:49] <burn> (ack == I am in line for you)
[05:40:27] <Christian Hoene> MIC: I can add a conformance test for PEAQ
[05:40:44] <Christian Hoene> with a few weeks
[05:40:50] <Christian Hoene> within a few weeks
[05:41:27] sun leaves the room
[05:41:28] Hendrik Scholz joins the room
[05:42:00] <Christian Hoene> MIC: To Erik, I do not agree
[05:42:26] <burn> ack Christian
[05:42:56] <Erik Norvell> Christian, which part do you not agree on?
[05:43:06] <Christian Hoene> Ha ha
[05:43:33] Jan Skoglund joins the room
[05:43:38] <Christian Hoene> Erik: Reducing the quality of decoder
[05:44:10] <Erik Norvell> Ok
[05:44:21] <burn> Christian, Erik: I can relay your conversation to each other at the mic if you want, or not — just let me know
[05:44:27] <Monty Montgomery> The problem is not 'allowing/refusing innovation', it's managing potential confusion.
[05:45:06] <Monty Montgomery> "The consumer needs to be sure what the certification actually means"
[05:45:26] <gmaxwell> I don't agree that it's okay to be loose in the decoder, we adopt a strict decoder in part to allow the encoder to run right to the rails. I can also speed from expirence with other formats (both Theora and Vorbis) that we've had significant difficulty convincing implementers to actually fix broken code even when there was no justification for the breakage but manpower.
[05:45:50] <gmaxwell> (and in particular the LP mode can easily be driven into loud screeching if there is mismatch between the encoder and decoder)
[05:45:50] <bemasc> mic: (this is a repeat) Request for clarification from the chairs: can the boundaries be loosened after moving to PROPOSED, if the requirements prove to be too tight in practice?
[05:45:58] mph joins the room
[05:46:40] <Christian Hoene> MIC: PEAQ has been verified for years. The other tool has not be standardized nor tested
[05:46:50] <Christian Hoene> by third parties
[05:46:50] mph leaves the room
[05:47:02] <burn> ack Christian
[05:47:12] <gmaxwell> MIC: The current configuration of how tight it was was a result of adjusting it so that it failed _real_ bugs in a second implementation. We do not know if its even possible to do that with PEAQ.
[05:47:13] <burn> Stephan coming up to mic to answer with his opinion as well
[05:47:14] hta joins the room
[05:47:25] <burn> ack gmaxwell. what's your name?
[05:47:31] <gmaxwell> Gregory Maxwell
[05:47:34] <burn> thx
[05:47:57] <burn> i'm next in line after Stephan
[05:48:03] <burn> fyi
[05:48:05] <gmaxwell> It was never mentioned, but thats where the current sensitivity comes from.
[05:48:26] <burn> gmaxwell: add to your mic comment?
[05:48:30] <gmaxwell> yes
[05:48:32] <gmaxwell> Thanks.
[05:48:34] <burn> ok
[05:48:54] janskoglund joins the room
[05:49:13] janskoglund leaves the room
[05:49:50] <Erik Norvell> MIC: Greg, the tools should be for verifying conformance, not catching bugs.
[05:50:08] <burn> ack Erik
[05:50:13] <burn> (after gmaxwell)
[05:50:40] <burn> Stephan's reply seemed germane, but monitoring to make sure it doesn't digress
[05:50:57] <gmaxwell> MIC: Yes, this was a second implementation which I would not have considered conformant due to errors. (I'm not saying I'm completely opposed to changing the threshold, but that was the factor)
[05:52:39] <burn> ack gmaxwell
[05:53:04] <gmaxwell> I'd be inclined to take this discussion to the list... I don't think we're making a lot of progress here.
[05:53:06] <gmaxwell> :)
[05:54:48] <bemasc> mic: I propose super-narrow.
[05:55:38] <bemasc> The value of a super-narrow test is to ensure that encoders can actually be optimized to a known objective function.
[05:55:50] <burn> ack ben
[05:56:24] <bemasc> Moreover: a too narrow test can always be broadened, but a too wide test cannot be narrowed.
[05:58:01] <burn> Ben, I think Matthew's question was directed at you. just waiting for your reply
[05:58:04] <bemasc> The value of supporting unusual use cases seems much lower to me than the problem of making it impossible to optimize encoders. How am I supposed to optimize an encoder if all the decoders sound different?
[05:58:27] <gmaxwell> Speaking from expirence we had problems for both Theora and Vorbis where implementors would not fix quality impacting bugs simply because fixing took man hours. (not anything to do with optimizing the decoder)
[05:58:40] <bemasc> The chair does not understand.
[06:00:18] <gmaxwell> Yea, if you want to relax it by 1 dB I wouldn't care. Allowing the 6502 would allow all bugs, it would allow embrace and extend, it would prevent encoders from driving the LPC filter near resonance (because a lower quality encoder would break people's ears)
[06:00:49] <rillian> lower quality *decoder*, gmaxwell meant
[06:00:54] <gmaxwell> er because a lower quality decoder in the last case.
[06:01:22] <Christian Hoene> MIC: Maybe two tests are needed. One for legal compliance with wide testing. A more narrow tests for interoperability.
[06:02:00] <burn> Cullen did not cut the line :)
[06:02:17] <Matthew Kaufman> chair fail
[06:02:33] Randell Jesup joins the room
[06:02:35] <burn> (now line is empty)
[06:02:45] <rillian> hi Randell
[06:02:49] <burn> waiting to relay for hum
[06:03:07] <Erik Norvell> humming for wide
[06:03:28] <rillian> mic: what do we consider the test in the current draft?
[06:03:33] <rillian> wide or narrow?
[06:03:39] <burn> who is rillian?
[06:03:42] <burn> sorry
[06:03:48] <rillian> Ralph Giles
[06:04:14] <burn> thx Ralph
[06:04:14] <Randell Jesup> Hi. sorry I'm late. Switch under my desk failed
[06:04:26] <burn> ack Erik for wide
[06:04:26] <Christian Hoene> MIC: Can we hum for both wide and narrow?
[06:04:28] <rillian> thanks
[06:04:35] <Christian Hoene> humm
[06:05:12] bemasc humm
[06:05:13] <gmaxwell> humming for narrow.
[06:05:13] <Christian Hoene> humm
[06:05:17] <rillian> humm for narrow
[06:05:19] <katwalsh> hum
[06:05:22] <Randell Jesup> humm
[06:05:29] <irc-> [ron_] hum narrow
[06:07:02] <Simon Romano> off-list (@Erik): btw, were you able to dial-in today?
[06:07:13] <Simon Romano> Slide 9: PLC With Real Traces (last meeting)
[06:07:14] <burn> back to slides
[06:07:43] <Simon Romano> Presentation stopped
[06:08:33] <burn> Paul Coverdale at mic
[06:08:45] <Erik Norvell> Simon: I got the stream working so I did not need dial-in. Thanks anyway!
[06:08:54] <Simon Romano> Great!I
[06:08:56] <Christian Hoene> I would like to see them on meetecho
[06:09:05] Hendrik Scholz leaves the room
[06:09:05] <Simon Romano> BTW, I do not have these slides on Meetecho..
[06:09:16] <Simon Romano> They were not on the meeting materials page, sorry.
[06:09:21] <bemasc> having them as a PDF on http would be worth something even if not on meetecho
[06:09:31] <rillian> worth quite a bit
[06:09:32] <burn> Tim will start talking while Cullen uploads
[06:09:34] <Matthew Kaufman> cullen is working on this
[06:09:40] <Matthew Kaufman> they should be up shortly
[06:09:42] <Simon Romano> OK...trying to recover now
[06:10:18] Hendrik Scholz joins the room
[06:10:28] <Simon Romano> Slide 1: Opus Testing
[06:10:32] <Simon Romano> Slide 3: Why we need more than formal listening t
[06:10:33] <Christian Hoene> Thanks!
[06:10:37] <Simon Romano> Here we go!
[06:10:46] rillian cheers
[06:12:23] victor.pascual.avila leaves the room
[06:12:51] <Simon Romano> Slide 4: The spec is software
[06:12:54] <Simon Romano> Slide 5: Continuous Integration
[06:12:59] <bemasc> are the slides up somewhere?
[06:13:12] <Simon Romano> Slide 6: Software Reliability Toolbox
[06:13:20] <burn> Simon, do you know if the slides are up?
[06:13:28] <burn> or anyone else who can check quickly?
[06:13:30] <Simon Romano> I think they're not yet up
[06:13:34] <gmaxwell> bemasc: see the links here... there was a delay at startup. They aren't on the website yet.
[06:13:40] <Matthew Kaufman> prob around lisde 7
[06:13:44] <Simon Romano> Cullen gave me a usb stick and I loaded them on Meetecho.
[06:14:05] <Simon Romano> Slide 7: Force Multipliers
[06:14:31] bemasc2 joins the room
[06:15:44] <Simon Romano> Slide 8: Operational Testing
[06:16:01] <bemasc> just so everyone knows, the slides are live in meetecho
[06:16:21] <burn> Meetecho rocks. it really simplifies jabber scribing :)
[06:16:22] tomkrist joins the room
[06:16:31] <burn> (as in, no need to scribe)
[06:16:36] <Simon Romano> Thanx ;-)
[06:16:45] <Simon Romano> Slide 9: Objective Quality Testing
[06:17:12] <Simon Romano> This is a cell phone ringing in the pocket of the speaker :-)
[06:17:22] <Christian Hoene> ha ha
[06:17:48] <burn> The cell phone is now long gone from the pocket, but we are getting some interesting background noise in the room
[06:17:49] <rillian> wow, that's annoying
[06:17:52] <Simon Romano> We're all listening to this...also those live in the room.
[06:17:52] <burn> over the speake
[06:18:03] <Simon Romano> Slide 10: Unit Tests
[06:18:21] <gmaxwell> (100 million frames of audio, but whats 100x between friends. ;) )
[06:18:38] <bemasc> heh, "small input space" of 4 billion options
[06:18:56] <Matthew Kaufman> dan, can you get magnus sitting next to you to see if he can track down AV person about this beeping
[06:19:15] tk joins the room
[06:19:19] <burn> magnus says yes
[06:19:34] <burn> he is working on it
[06:19:37] <Simon Romano> Slide 11: Static Analysis
[06:21:03] <burn> Tim switched to handheld mic
[06:21:05] <Simon Romano> OK! We switched off the speaker's mike...
[06:21:06] <burn> seems to be better
[06:21:08] <Simon Romano> Slide 12: Manual Instrumentation
[06:21:08] <gmaxwell> omg hurray!
[06:21:12] <rillian> thanks you
[06:21:41] <Randell Jesup> yeah!
[06:21:51] <Simon Romano> Slide 13: Automatic Instrumentation
[06:22:20] victor.pascual.avila joins the room
[06:22:32] <burn> FYI, magnus has told AV people about the wireless mic
[06:23:12] <Simon Romano> Slide 14: Line and Branch Coverage Analysis
[06:24:23] <Simon Romano> Slide 15: Decoder Fuzzing
[06:25:07] <Matthew Kaufman> wireless mic is simply incompatible with the current speaker
[06:25:48] <burn> Tim's EM field must be amazing
[06:26:00] Bernhard Feiten joins the room
[06:26:04] <rillian> It's the implants
[06:26:10] <Simon Romano> Slide 16: Whitebox Fuzzing
[06:26:14] <gmaxwell> Tim doesn't own a phone, now we know why.
[06:26:21] <Matthew Kaufman> @rillian: that's what i assumed
[06:26:31] <irc-> [xiphmont] Time has a loose fan belt.
[06:26:34] <irc-> [xiphmont] err, Tim.
[06:26:52] <Simon Romano> Slide 17: Encoder Fuzzing
[06:28:00] <Simon Romano> Slide 18: Multiplatform Testing
[06:29:12] <Simon Romano> Slide 19: Additional Testing
[06:29:49] <Simon Romano> Slide 20: Toolchain Bugs
[06:29:57] <burn> Jonathan Lennox at mic
[06:31:05] <gmaxwell> (random jabber chatter) I tried pdp-11 but I couldn't find a C compiler near enough to ANSI to not drive me insane..
[06:31:15] <Simon Romano> Slide 21: Implementation Interop Testing
[06:33:05] <Simon Romano> Slide 22: Implementation Interop Testing
[06:33:15] Hendrik Scholz leaves the room
[06:34:03] <Christian Hoene> MIC: Good work, impressive
[06:34:11] Hendrik Scholz joins the room
[06:34:15] <Simon Romano> Presentation stopped
[06:34:26] <Jan Skoglund> Agree, impressive
[06:35:08] <Simon Romano> Presentation stopped
[06:35:12] <Randell Jesup> ditto
[06:35:28] <burn> Tim at mic
[06:35:56] <Simon Romano> Slide 1: Issues Found During WGLC
[06:36:00] <Simon Romano> Slide 3: Issue #2
[06:36:13] victor.pascual.avila leaves the room
[06:36:36] <Simon Romano> Slide 4: Issue #3
[06:36:41] <Simon Romano> Slide 3: Issue #2
[06:36:42] <burn> Paul Coverdale at mic
[06:37:00] <burn> Tim Tb at mic
[06:37:18] <burn> Jean-Marc Valin at mic
[06:37:28] <Simon Romano> Slide 4: Issue #3
[06:37:34] <gmaxwell> We do have DTMF in my bulk test sets but not the automated ones, but I don't test a DTMF decoder. :)
[06:37:47] <Simon Romano> Slide 3: Issue #2
[06:38:23] victor.pascual.avila joins the room
[06:38:53] <Simon Romano> Slide 4: Issue #3
[06:38:54] <Simon Romano> Slide 5: Issue #4
[06:39:00] <Simon Romano> Slide 3: Issue #2
[06:39:00] <burn> jonathan lennox at mic
[06:39:26] <Simon Romano> Slide 4: Issue #3
[06:39:28] <Simon Romano> Slide 5: Issue #4
[06:39:30] <Simon Romano> Slide 4: Issue #3
[06:39:35] <Simon Romano> Slide 5: Issue #4
[06:40:57] <Simon Romano> Presentation stopped
[06:41:20] <burn> That was Jean-Marc at mic who said "yes"
[06:41:57] <Simon Romano> Slide 1: Opus Testing with Network Traces
[06:41:59] <Simon Romano> Slide 2: Tests with Network Traces
[06:42:35] Matt Kotvis leaves the room
[06:43:42] <Simon Romano> Slide 3: Results for Network Traces
[06:44:00] victor.pascual.avila leaves the room
[06:44:13] victor.pascual.avila joins the room
[06:44:41] <Simon Romano> Slide 4: Results for Random Packet Loss
[06:44:48] <Simon Romano> Slide 4: Results for Random Packet Loss
[06:45:18] <burn> Paul Coverdale at mic
[06:46:34] <burn> Jean-Marc Valin at mic
[06:48:03] <burn> Paul Coverdale at mic
[06:48:32] RjS leaves the room
[06:48:44] <Christian Hoene> MIC: DTMF is part of the requirements, modem or fax was excluded from the requirements doc
[06:48:47] <burn> Jean-Marc Valin at mic
[06:48:57] <burn> ack Christian
[06:48:58] <bemasc> The guidelines already express a consensus to rule these cases out.
[06:48:58] <gmaxwell> MIC: Lets just assume modems don't work. Does anyone consider that a blocker?
[06:49:09] <burn> ack gmaxwell
[06:49:16] bemasc is covered by hoene's comment
[06:49:21] <burn> I am second or third in line i think
[06:49:28] <burn> Brian rosen at mic
[06:49:32] <burn> I think I'm next
[06:49:58] <gmaxwell> Can brian test it?
[06:50:20] <Christian Hoene> Some tools are out there to test DTMF
[06:50:36] <gmaxwell> yea, dtmf isn't a big deal.. dunno about TTY
[06:50:43] <Christian Hoene> There has been a testing DTMF CD somewhere out here
[06:50:48] <Christian Hoene> I can look it up
[06:50:54] <Matthew Kaufman> mitel test tape!
[06:51:20] <Christian Hoene> yes, thanks!
[06:52:05] <irc-> [ron_] that's about testing DTMF detectors. not passthrough of codecs
[06:52:46] <burn> irc-, who are you and do you want that at the mic?
[06:52:54] <irc-> [ron_] and is mostly about false positive triggering on speech, not corrupting 'real' DTMF
[06:53:03] <Jonathan Lennox> Still, it means you're testing edge-of-spec DTMF, not just pure/correct DTMF.
[06:53:22] <irc-> [ron_] no, that's just comment for people here, we can do this on the list
[06:53:33] <burn> irc-: okay, thx
[06:53:43] <Randell Jesup> awww. I've run iLBC at 50% loss and could understand the audio. Wouldn't say it sounded nice, of course. ;-)
[06:53:52] <Matthew Kaufman> yeah, the mitel tests include anti-talkoff and dialtone reject but also a bunch of timing and twist edge cases
[06:54:02] <gmaxwell> Yea, we posted examples months ago with 50% loss that were intelligible. :)
[06:54:16] <gmaxwell> (where months == a year, actually)
[06:54:16] <Simon Romano> That's all folks!
See you soon on Meetecho ;-)
[06:54:18] <burn> meeting adjourned
[06:54:23] <gmaxwell> Thanks all!
[06:54:28] bemasc leaves the room
[06:54:32] <gmaxwell> (and thanks for relaying to the mic!)
[06:54:32] Hendrik Scholz leaves the room
[06:54:34] <rillian> thanks everyone
[06:54:35] <burn> my pleasure
[06:54:38] <Erik Norvell> thanks!
[06:54:40] <Christian Hoene> good work guys.
[06:54:41] <Randell Jesup> no cookies? :-)
[06:54:42] <burn> was a good discussion
[06:54:45] hta leaves the room
[06:54:50] <rillian> especially burns and simon for making remote attendance feasible!
[06:54:53] <Randell Jesup> Thanks everyone
[06:54:55] <Matthew Kaufman> i have a bunch of real-world DTMF-over-radio samples recorded over years at 22050 Hz and a decoder. i can test impairments on the path between those files and the decoder, but i don't know how useful that is.
[06:55:07] <Simon Romano> Our pleasure, really.
[06:55:17] Simon Romano leaves the room
[06:55:29] <gmaxwell> Matthew Kaufman: my collection contains about 5 minutes of dtmf.. so they've been tested with PEAQ and opus compare.. but not a dtmf decoder.
[06:55:30] Erik Norvell leaves the room
[06:56:02] tsuichi leaves the room
[06:56:05] burn leaves the room
[06:56:09] <Matthew Kaufman> offline for me
[06:56:20] <Christian Hoene> bye
[06:56:21] Christian Hoene leaves the room
[06:56:32] Matthew Kaufman leaves the room
[06:56:34] Monty Montgomery leaves the room
[06:56:37] <Randell Jesup> DTMF decoders have acceptance variances, so hard to quantify pass/fail
[06:56:48] <Randell Jesup> varies from decoder to decoder
[06:57:07] <Randell Jesup> so you'd need to test with a variety
[06:57:26] <rillian> the internet claims TTY tones are single frequency, switching between two tones
[06:57:47] <rillian> no wonder it's at 45.5 baud
[06:57:55] <Randell Jesup> Sounds right
[06:57:59] <gmaxwell> oh.. that kind of tty?
[06:58:02] <gmaxwell> I can decode that!
[06:58:05] <gmaxwell> I use that over HF.
[06:58:10] <rillian> oh? cool
[06:58:16] <gmaxwell> sure sure. I'll test it.
[06:58:18] <Randell Jesup> TTY sucks dead gerbils through a garden hose
[06:58:18] <rillian> wikipedia doesn't actually describe the encoding
[06:58:29] Lorenzo Miniero leaves the room
[06:58:48] <Randell Jesup> (Old Amiga saying)
[06:58:53] <rillian> but yes, I believe Brian meant text-over-POTS devices for the deaf
[06:58:53] RjS joins the room
[06:59:08] <gmaxwell>
[06:59:08] <rillian> Randell Jesup: it sounded familiar
[06:59:26] RjS leaves the room
[06:59:46] <gmaxwell> 45 baud tty is used on HF (well not as much as other modes anymore)
[07:00:17] <rillian> gmaxwell: "RTTY"?
[07:00:19] <Jonathan Rosenberg> ++
[07:00:45] Randell Jesup leaves the room
[07:00:59] <rillian> 45 is the original telegraph-powered teletype rate, right?
[07:01:10] Pen leaves the room: Computer went to sleep
[07:01:20] tk leaves the room
[07:01:21] <gmaxwell> Yea, I'm pretty sure it's just called RTTY when its over radio, but I'll check to make sure its the same.
[07:01:24] Jonathan Rosenberg leaves the room
[07:01:27] Jan Skoglund leaves the room
[07:01:35] Bernhard Feiten leaves the room
[07:01:38] Ralph Giles leaves the room
[07:01:47] MeetechoAudioChannel leaves the room
[07:01:52] bemasc2 leaves the room
[07:02:24] <rillian> probably at least related
[07:02:39] <gmaxwell> Closing IRC bridge, thanks whoever was there (ron!)
[07:02:41] <rillian> apparently there were a number of regional standards
[07:02:48] <rillian> gmaxwell: thanks for running that, gmaxwell
[07:03:14] <gmaxwell> rillian: the internet says it's just some set of carrier and tone spacings. (but thats unsurprising, it's not uniform on HF either)
[07:03:26] irc- leaves the room
[07:03:51] katwalsh leaves the room
[07:04:27] <rillian> well, my 'two pure tones' is based entirely on <>
[07:04:39] <rillian> oops, typo
[07:04:52] <rillian> <>
[07:05:50] <rillian> gated carrier makes more sense given the age
[07:06:11] <rillian> but maybe they need a continuous sound for keep-alive on a circuit-switched network?
[07:06:29] <gmaxwell> nah, it's frequency shift.. by carrier I just mean the sound of zero that gets shifted.
[07:06:46] <rillian> ah, ok
[07:07:05] <gmaxwell> In anycase, I don't anticipate a problem since we've run plenty of synthetic signals that look just like that.
[07:07:13] <rillian> *nod*
[07:08:19] gmaxwell back to irc
[07:08:20] gmaxwell leaves the room
[07:09:10] RjS joins the room
[07:09:32] pcgod leaves the room
[07:20:09] Matthew Kaufman joins the room
[07:22:24] Jonathan Lennox leaves the room
[07:24:32] rillian leaves the room
[07:49:18] RjS leaves the room
[07:50:41] tomkrist leaves the room
[07:55:33] Matthew Kaufman leaves the room
[08:04:36] victor.pascual.avila leaves the room
[08:32:34] tsuichi joins the room
[08:54:22] tsuichi leaves the room
[08:56:54] victor.pascual.avila joins the room
[09:41:39] Matthew Kaufman joins the room
[09:42:03] victor.pascual.avila leaves the room
[10:12:48] burn joins the room
[10:26:33] Matthew Kaufman leaves the room
[11:17:14] burn leaves the room
[17:09:19] mjgraves joins the room
[17:09:19] mjgraves leaves the room