rpms/opus
7
0
Fork 0
Code Issues Pull Requests Releases Activity
ce65dcc667
opus/draft-ietf-codec-opus-update-04
Peter Robinson ce65dcc667 1.2 alpha
2016-11-04 21:33:58 +00:00

754 lines
33 KiB
Plaintext
Raw Blame History

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
<head profile="http://dublincore.org/documents/2008/08/04/dc-html/">
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<meta name="robots" content="index,follow" />
<meta name="creator" content="rfcmarkup version 1.120" />
<link rel="schema.DC" href="http://purl.org/dc/elements/1.1/" />
<meta name="DC.Relation.Replaces" content="draft-valin-codec-opus-update" />
<meta name="DC.Identifier" content="urn:ietf:id:ietf-codec-opus-update" />
<meta name="DC.Date.Issued" content="2016-10-21" />
<meta name="DC.Creator" content="Valin, Jean-Marc" />
<meta name="DC.Creator" content="Vos, Koen" />
<meta name="DC.Description.Abstract" content="This document addresses minor issues that were found in the
specification of the Opus audio codec in RFC 6716 [RFC6716]." />
<meta name="DC.Title" content="Updates to the Opus Audio Codec" />
<link rel="icon" href="/images/id.png" type="image/png" />
<link rel="shortcut icon" href="/images/id.png" type="image/png" />
<title>draft-ietf-codec-opus-update-04 - Updates to the Opus Audio Codec</title>
<style type="text/css">
@media only screen
and (min-width: 992px)
and (max-width: 1199px) {
body { font-size: 14pt; }
div.content { width: 96ex; margin: 0 auto; }
}
@media only screen
and (min-width: 768px)
and (max-width: 991px) {
body { font-size: 14pt; }
div.content { width: 96ex; margin: 0 auto; }
}
@media only screen
and (min-width: 480px)
and (max-width: 767px) {
body { font-size: 11pt; }
div.content { width: 96ex; margin: 0 auto; }
}
@media only screen
and (max-width: 479px) {
body { font-size: 8pt; }
div.content { width: 96ex; margin: 0 auto; }
}
@media only screen
and (min-device-width : 375px)
and (max-device-width : 667px) {
body { font-size: 9.5pt; }
div.content { width: 96ex; margin: 0 1px; }
}
@media only screen
and (min-device-width: 1200px) {
body { font-size: 10pt; margin: 0 4em; }
div.content { width: 96ex; margin: 0; }
}
h1, h2, h3, h4, h5, h6, .h1, .h2, .h3, .h4, .h5, .h6 {
font-weight: bold;
line-height: 0pt;
display: inline;
white-space: pre;
font-family: monospace;
font-size: 1em;
font-weight: bold;
}
pre {
font-size: 1em;
margin-top: 0px;
margin-bottom: 0px;
}
.pre {
white-space: pre;
font-family: monospace;
}
.header{
font-weight: bold;
}
.newpage {
page-break-before: always;
}
.invisible {
text-decoration: none;
color: white;
}
a.selflink {
color: black;
text-decoration: none;
}
@media print {
body {
font-family: monospace;
font-size: 10.5pt;
}
h1, h2, h3, h4, h5, h6 {
font-size: 1em;
}
a:link, a:visited {
color: inherit;
text-decoration: none;
}
.noprint {
display: none;
}
}
@media screen {
.grey, .grey a:link, .grey a:visited {
color: #777;
}
.docinfo {
background-color: #EEE;
}
.top {
border-top: 7px solid #EEE;
}
.bgwhite { background-color: white; }
.bgred { background-color: #F44; }
.bggrey { background-color: #666; }
.bgbrown { background-color: #840; }
.bgorange { background-color: #FA0; }
.bgyellow { background-color: #EE0; }
.bgmagenta{ background-color: #F4F; }
.bgblue { background-color: #66F; }
.bgcyan { background-color: #4DD; }
.bggreen { background-color: #4F4; }
.legend { font-size: 90%; }
.cplate { font-size: 70%; border: solid grey 1px; }
}
</style>
<!--[if IE]>
<style>
body {
font-size: 13px;
margin: 10px 10px;
}
</style>
<![endif]-->
<script type="text/javascript"><!--
function addHeaderTags() {
var spans = document.getElementsByTagName("span");
for (var i=0; i < spans.length; i++) {
var elem = spans[i];
if (elem) {
var level = elem.getAttribute("class");
if (level == "h1" || level == "h2" || level == "h3" || level == "h4" || level == "h5" || level == "h6") {
elem.innerHTML = "<"+level+">"+elem.innerHTML+"</"+level+">";
}
}
}
}
var legend_html = "Colour legend:<br /> <table> <tr><td>Unknown:</td> <td><span class='cplate bgwhite'>&nbsp;&nbsp;&nbsp;&nbsp;</span></td></tr> <tr><td>Draft:</td> <td><span class='cplate bgred'>&nbsp;&nbsp;&nbsp;&nbsp;</span></td></tr> <tr><td>Informational:</td> <td><span class='cplate bgorange'>&nbsp;&nbsp;&nbsp;&nbsp;</span></td></tr> <tr><td>Experimental:</td> <td><span class='cplate bgyellow'>&nbsp;&nbsp;&nbsp;&nbsp;</span></td></tr> <tr><td>Best Common Practice:</td> <td><span class='cplate bgmagenta'>&nbsp;&nbsp;&nbsp;&nbsp;</span></td></tr> <tr><td>Proposed Standard:</td> <td><span class='cplate bgblue'>&nbsp;&nbsp;&nbsp;&nbsp;</span></td></tr> <tr><td>Draft Standard (old designation):</td> <td><span class='cplate bgcyan'>&nbsp;&nbsp;&nbsp;&nbsp;</span></td></tr> <tr><td>Internet Standard:</td> <td><span class='cplate bggreen'>&nbsp;&nbsp;&nbsp;&nbsp;</span></td></tr> <tr><td>Historic:</td> <td><span class='cplate bggrey'>&nbsp;&nbsp;&nbsp;&nbsp;</span></td></tr> <tr><td>Obsolete:</td> <td><span class='cplate bgbrown'>&nbsp;&nbsp;&nbsp;&nbsp;</span></td></tr> </table>";
function showElem(id) {
var elem = document.getElementById(id);
elem.innerHTML = eval(id+"_html");
elem.style.visibility='visible';
}
function hideElem(id) {
var elem = document.getElementById(id);
elem.style.visibility='hidden';
elem.innerHTML = "";
}
// -->
</script>
</head>
<body onload="addHeaderTags()">
<div class="content">
<div style="height: 13px;">
<div onmouseover="this.style.cursor='pointer';"
onclick="showElem('legend');"
onmouseout="hideElem('legend')"
style="height: 6px; position: absolute;"
class="pre noprint docinfo bgred"
title="Click for colour legend." > </div>
<div id="legend"
class="docinfo noprint pre legend"
style="position:absolute; top: 4px; left: 4ex; visibility:hidden; background-color: white; padding: 4px 9px 5px 7px; border: solid #345 1px; "
onmouseover="showElem('legend');"
onmouseout="hideElem('legend');">
</div>
</div>
<span class="pre noprint docinfo top">[<a href="../html/" title="Document search and retrieval page">Docs</a>] [<a href="https://tools.ietf.org/id/draft-ietf-codec-opus-update-04.txt" title="Plaintext version of this document">txt</a>|<a href="/pdf/draft-ietf-codec-opus-update-04.txt" title="PDF version of this document">pdf</a>] [<a href='https://datatracker.ietf.org/doc/draft-ietf-codec-opus-update' title='IESG Datatracker information for this document'>Tracker</a>] [<a href="../wg/codec" title="The working group handling this document">WG</a>] [<a href="mailto:draft-ietf-codec-opus-update@tools.ietf.org?subject=draft-ietf-codec-opus-update%20" title="Send email to the document authors">Email</a>] [<a href="/rfcdiff?difftype=--hwdiff&amp;url2=draft-ietf-codec-opus-update-04.txt" title="Inline diff (wdiff)">Diff1</a>] [<a href="/rfcdiff?url2=draft-ietf-codec-opus-update-04.txt" title="Side-by-side diff">Diff2</a>] [<a href="/idnits?url=https://tools.ietf.org/id/draft-ietf-codec-opus-update-04.txt" title="Run an idnits check of this document">Nits</a>] </span><br />
<span class="pre noprint docinfo"> </span><br />
<span class="pre noprint docinfo">Versions: (<a href="./draft-valin-codec-opus-update" title="Precursor">draft-valin-codec-opus-update</a>) <a href="./draft-ietf-codec-opus-update-00">00</a>
<a href="./draft-ietf-codec-opus-update-01">01</a> <a href="./draft-ietf-codec-opus-update-02">02</a> <a href="./draft-ietf-codec-opus-update-03">03</a> <a href="./draft-ietf-codec-opus-update-04">04</a> </span><br />
<span class="pre noprint docinfo"> </span><br />
<pre>
Network Working Group JM. Valin
Internet-Draft Mozilla Corporation
Intended status: Standards Track K. Vos
Expires: April 24, 2017 vocTone
October 21, 2016
<span class="h1">Updates to the Opus Audio Codec</span>
<span class="h1">draft-ietf-codec-opus-update-04</span>
Abstract
This document addresses minor issues that were found in the
specification of the Opus audio codec in <a href="./rfc6716">RFC 6716</a> [<a href="./rfc6716" title="&quot;Definition of the Opus Audio Codec&quot;">RFC6716</a>].
Status of This Memo
This Internet-Draft is submitted in full conformance with the
provisions of <a href="./bcp78">BCP 78</a> and <a href="./bcp79">BCP 79</a>.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF). Note that other groups may also distribute
working documents as Internet-Drafts. The list of current Internet-
Drafts is at <a href="http://datatracker.ietf.org/drafts/current/">http://datatracker.ietf.org/drafts/current/</a>.
Internet-Drafts are draft documents valid for a maximum of six months
and may be updated, replaced, or obsoleted by other documents at any
time. It is inappropriate to use Internet-Drafts as reference
material or to cite them other than as "work in progress."
This Internet-Draft will expire on April 24, 2017.
Copyright Notice
Copyright (c) 2016 IETF Trust and the persons identified as the
document authors. All rights reserved.
This document is subject to <a href="./bcp78">BCP 78</a> and the IETF Trust's Legal
Provisions Relating to IETF Documents
(<a href="http://trustee.ietf.org/license-info">http://trustee.ietf.org/license-info</a>) in effect on the date of
publication of this document. Please review these documents
carefully, as they describe your rights and restrictions with respect
to this document. Code Components extracted from this document must
include Simplified BSD License text as described in Section 4.e of
the Trust Legal Provisions and are provided without warranty as
described in the Simplified BSD License.
<span class="grey">Valin &amp; Vos Expires April 24, 2017 [Page 1]</span></pre>
<hr class='noprint' style='width: 96ex;' align='left'/><!--NewPage--><pre class='newpage'><a name="page-2" id="page-2" href="#page-2" class="invisible"> </a>
<span class="grey">Internet-Draft Opus Update October 2016</span>
Table of Contents
<a href="#section-1">1</a>. Introduction . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-2">2</a>
<a href="#section-2">2</a>. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-2">2</a>
<a href="#section-3">3</a>. Stereo State Reset in SILK . . . . . . . . . . . . . . . . . <a href="#page-2">2</a>
<a href="#section-4">4</a>. Parsing of the Opus Packet Padding . . . . . . . . . . . . . <a href="#page-3">3</a>
<a href="#section-5">5</a>. Resampler buffer . . . . . . . . . . . . . . . . . . . . . . <a href="#page-4">4</a>
<a href="#section-6">6</a>. Integer wrap-around in inverse gain computation . . . . . . . <a href="#page-5">5</a>
<a href="#section-7">7</a>. Integer wrap-around in LSF decoding . . . . . . . . . . . . . <a href="#page-6">6</a>
<a href="#section-8">8</a>. Cap on Band Energy . . . . . . . . . . . . . . . . . . . . . <a href="#page-6">6</a>
<a href="#section-9">9</a>. Hybrid Folding . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-7">7</a>
<a href="#section-10">10</a>. Downmix to Mono . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-9">9</a>
<a href="#section-11">11</a>. New Test Vectors . . . . . . . . . . . . . . . . . . . . . . <a href="#page-9">9</a>
<a href="#section-12">12</a>. IANA Considerations . . . . . . . . . . . . . . . . . . . . . <a href="#page-9">9</a>
<a href="#section-13">13</a>. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . <a href="#page-9">9</a>
<a href="#section-14">14</a>. References . . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-10">10</a>
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-10">10</a>
<span class="h2"><a class="selflink" name="section-1" href="#section-1">1</a>. Introduction</span>
This document addresses minor issues that were discovered in the
reference implementation of the Opus codec that serves as the
specification in <a href="./rfc6716">RFC 6716</a> [<a href="./rfc6716" title="&quot;Definition of the Opus Audio Codec&quot;">RFC6716</a>]. Only issues affecting the
decoder are listed here. An up-to-date implementation of the Opus
encoder can be found at <a href="http://opus-codec.org/">http://opus-codec.org/</a>. The updated
specification remains fully compatible with the original
specification.
<span class="h2"><a class="selflink" name="section-2" href="#section-2">2</a>. Terminology</span>
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be interpreted as described in <a href="./rfc2119">RFC 2119</a> [<a href="./rfc2119" title="&quot;Key words for use in RFCs to Indicate Requirement Levels&quot;">RFC2119</a>].
<span class="h2"><a class="selflink" name="section-3" href="#section-3">3</a>. Stereo State Reset in SILK</span>
The reference implementation does not reinitialize the stereo state
during a mode switch. The old stereo memory can produce a brief
impulse (i.e. single sample) in the decoded audio. This can be fixed
by changing silk/dec_API.c at line 72:
<span class="grey">Valin &amp; Vos Expires April 24, 2017 [Page 2]</span></pre>
<hr class='noprint' style='width: 96ex;' align='left'/><!--NewPage--><pre class='newpage'><a name="page-3" id="page-3" href="#page-3" class="invisible"> </a>
<span class="grey">Internet-Draft Opus Update October 2016</span>
for( n = 0; n &lt; DECODER_NUM_CHANNELS; n++ ) {
ret = silk_init_decoder( &amp;channel_state[ n ] );
}
+ silk_memset(&amp;((silk_decoder *)decState)-&gt;sStereo, 0,
+ sizeof(((silk_decoder *)decState)-&gt;sStereo));
+ /* Not strictly needed, but it's cleaner that way */
+ ((silk_decoder *)decState)-&gt;prev_decode_only_middle = 0;
return ret;
}
This change affects the normative part of the decoder, although the
amount of change is too small to make a significant impact on
testvectors.
<span class="h2"><a class="selflink" name="section-4" href="#section-4">4</a>. Parsing of the Opus Packet Padding</span>
It was discovered that some invalid packets of very large size could
trigger an out-of-bounds read in the Opus packet parsing code
responsible for padding. This is due to an integer overflow if the
signaled padding exceeds 2^31-1 bytes (the actual packet may be
smaller). The code can be fixed by applying the following changes at
line 596 of src/opus_decoder.c:
/* Padding flag is bit 6 */
if (ch&amp;0x40)
{
- int padding=0;
int p;
do {
if (len&lt;=0)
return OPUS_INVALID_PACKET;
p = *data++;
len--;
- padding += p==255 ? 254: p;
+ len -= p==255 ? 254: p;
} while (p==255);
- len -= padding;
}
This packet parsing issue is limited to reading memory up to about 60
kB beyond the compressed buffer. This can only be triggered by a
compressed packet more than about 16 MB long, so it's not a problem
for RTP. In theory, it _could_ crash a file decoder (e.g. Opus in
Ogg) if the memory just after the incoming packet is out-of-range,
but our attempts to trigger such a crash in a production application
built using an affected version of the Opus decoder failed.
<span class="grey">Valin &amp; Vos Expires April 24, 2017 [Page 3]</span></pre>
<hr class='noprint' style='width: 96ex;' align='left'/><!--NewPage--><pre class='newpage'><a name="page-4" id="page-4" href="#page-4" class="invisible"> </a>
<span class="grey">Internet-Draft Opus Update October 2016</span>
<span class="h2"><a class="selflink" name="section-5" href="#section-5">5</a>. Resampler buffer</span>
The SILK resampler had the following issues:
1. The calls to memcpy() were using sizeof(opus_int32), but the type
of the local buffer was opus_int16.
2. Because the size was wrong, this potentially allowed the source
and destination regions of the memcpy() to overlap. We _believe_
that nSamplesIn is at least fs_in_khZ, which is at least 8.
Since RESAMPLER_ORDER_FIR_12 is only 8, that should not be a
problem once the type size is fixed.
3. The size of the buffer used RESAMPLER_MAX_BATCH_SIZE_IN, but the
data stored in it was actually _twice_ the input batch size
(nSamplesIn&lt;&lt;1).
The fact that the code never produced any error in testing (including
when run under the Valgrind memory debugger), suggests that in
practice the batch sizes are reasonable enough that none of the
issues above was ever a problem. However, proving that is non-
obvious.
The code can be fixed by applying the following changes to line 70 of
silk/resampler_private_IIR_FIR.c:
)
{
silk_resampler_state_struct *S = \
(silk_resampler_state_struct *)SS;
opus_int32 nSamplesIn;
opus_int32 max_index_Q16, index_increment_Q16;
- opus_int16 buf[ RESAMPLER_MAX_BATCH_SIZE_IN + \
RESAMPLER_ORDER_FIR_12 ];
+ opus_int16 buf[ 2*RESAMPLER_MAX_BATCH_SIZE_IN + \
RESAMPLER_ORDER_FIR_12 ];
/* Copy buffered samples to start of buffer */
- silk_memcpy( buf, S-&gt;sFIR, RESAMPLER_ORDER_FIR_12 \
* sizeof( opus_int32 ) );
+ silk_memcpy( buf, S-&gt;sFIR, RESAMPLER_ORDER_FIR_12 \
* sizeof( opus_int16 ) );
/* Iterate over blocks of frameSizeIn input samples */
index_increment_Q16 = S-&gt;invRatio_Q16;
while( 1 ) {
nSamplesIn = silk_min( inLen, S-&gt;batchSize );
<span class="grey">Valin &amp; Vos Expires April 24, 2017 [Page 4]</span></pre>
<hr class='noprint' style='width: 96ex;' align='left'/><!--NewPage--><pre class='newpage'><a name="page-5" id="page-5" href="#page-5" class="invisible"> </a>
<span class="grey">Internet-Draft Opus Update October 2016</span>
/* Upsample 2x */
silk_resampler_private_up2_HQ( S-&gt;sIIR, &amp;buf[ \
RESAMPLER_ORDER_FIR_12 ], in, nSamplesIn );
max_index_Q16 = silk_LSHIFT32( nSamplesIn, 16 + 1 \
); /* + 1 because 2x upsampling */
out = silk_resampler_private_IIR_FIR_INTERPOL( out, \
buf, max_index_Q16, index_increment_Q16 );
in += nSamplesIn;
inLen -= nSamplesIn;
if( inLen &gt; 0 ) {
/* More iterations to do; copy last part of \
filtered signal to beginning of buffer */
- silk_memcpy( buf, &amp;buf[ nSamplesIn &lt;&lt; 1 ], \
RESAMPLER_ORDER_FIR_12 * sizeof( opus_int32 ) );
+ silk_memmove( buf, &amp;buf[ nSamplesIn &lt;&lt; 1 ], \
RESAMPLER_ORDER_FIR_12 * sizeof( opus_int16 ) );
} else {
break;
}
}
/* Copy last part of filtered signal to the state for \
the next call */
- silk_memcpy( S-&gt;sFIR, &amp;buf[ nSamplesIn &lt;&lt; 1 ], \
RESAMPLER_ORDER_FIR_12 * sizeof( opus_int32 ) );
+ silk_memcpy( S-&gt;sFIR, &amp;buf[ nSamplesIn &lt;&lt; 1 ], \
RESAMPLER_ORDER_FIR_12 * sizeof( opus_int16 ) );
}
Note: due to RFC formatting conventions, lines exceeding the column
width in the patch above are split using a backslash character. The
backslashes at the end of a line and the white space at the beginning
of the following line are not part of the patch. A properly
formatted patch including the three changes above is available at
&lt;<a href="https://jmvalin.ca/misc_stuff/opus_update.patch">https://jmvalin.ca/misc_stuff/opus_update.patch</a>&gt;. (EDITOR: change
to an ietf.org link when ready)
<span class="h2"><a class="selflink" name="section-6" href="#section-6">6</a>. Integer wrap-around in inverse gain computation</span>
It was discovered through decoder fuzzing that some bitstreams could
produce integer values exceeding 32-bits in
LPC_inverse_pred_gain_QA(), causing a wrap-around. Although the
error is harmless in practice, the C standard considers the behavior
as undefined, so the following patch to line 87 of silk/
LPC_inv_pred_gain.c detects values that do not fit in a 32-bit
integer and considers the corresponding filters unstable:
<span class="grey">Valin &amp; Vos Expires April 24, 2017 [Page 5]</span></pre>
<hr class='noprint' style='width: 96ex;' align='left'/><!--NewPage--><pre class='newpage'><a name="page-6" id="page-6" href="#page-6" class="invisible"> </a>
<span class="grey">Internet-Draft Opus Update October 2016</span>
/* Update AR coefficient */
for( n = 0; n &lt; k; n++ ) {
- tmp_QA = Aold_QA[ n ] - MUL32_FRAC_Q( \
Aold_QA[ k - n - 1 ], rc_Q31, 31 );
- Anew_QA[ n ] = MUL32_FRAC_Q( tmp_QA, rc_mult2 , mult2Q );
+ opus_int64 tmp64;
+ tmp_QA = silk_SUB_SAT32( Aold_QA[ n ], MUL32_FRAC_Q( \
Aold_QA[ k - n - 1 ], rc_Q31, 31 ) );
+ tmp64 = silk_RSHIFT_ROUND64( silk_SMULL( tmp_QA, \
rc_mult2 ), mult2Q);
+ if( tmp64 &gt; silk_int32_MAX || tmp64 &lt; silk_int32_MIN ) {
+ return 0;
+ }
+ Anew_QA[ n ] = ( opus_int32 )tmp64;
}
<span class="h2"><a class="selflink" name="section-7" href="#section-7">7</a>. Integer wrap-around in LSF decoding</span>
It was discovered -- also from decoder fuzzing -- that an integer
wrap-around could occur when decoding line spectral frequency
coefficients from extreme bitstreams. The end result of the wrap-
around is an illegal read access on the stack, which the authors do
not believe is exploitable but should nonetheless be fixed. The
following patch to line 137 of silk/NLSF_stabilize.c prevents the
problem:
/* Keep delta_min distance between the NLSFs */
for( i = 1; i &lt; L; i++ )
- NLSF_Q15[i] = silk_max_int( NLSF_Q15[i], \
NLSF_Q15[i-1] + NDeltaMin_Q15[i] );
+ NLSF_Q15[i] = silk_max_int( NLSF_Q15[i], \
silk_ADD_SAT16( NLSF_Q15[i-1], NDeltaMin_Q15[i] ) );
/* Last NLSF should be no higher than 1 - NDeltaMin[L] */
<span class="h2"><a class="selflink" name="section-8" href="#section-8">8</a>. Cap on Band Energy</span>
On extreme bit-streams, it is possible for log-domain band energy
levels to exceed the maximum single-precision floating point value
once converted to a linear scale. This would later cause the decoded
values to be NaN, possibly causing problems in the software using the
PCM values. This can be avoided with the following patch to line 552
of celt/quant_bands.c:
<span class="grey">Valin &amp; Vos Expires April 24, 2017 [Page 6]</span></pre>
<hr class='noprint' style='width: 96ex;' align='left'/><!--NewPage--><pre class='newpage'><a name="page-7" id="page-7" href="#page-7" class="invisible"> </a>
<span class="grey">Internet-Draft Opus Update October 2016</span>
{
opus_val16 lg = ADD16(oldEBands[i+c*m-&gt;nbEBands],
SHL16((opus_val16)eMeans[i],6));
+ lg = MIN32(QCONST32(32.f, 16), lg);
eBands[i+c*m-&gt;nbEBands] = PSHR32(celt_exp2(lg),4);
}
for (;i&lt;m-&gt;nbEBands;i++)
<span class="h2"><a class="selflink" name="section-9" href="#section-9">9</a>. Hybrid Folding</span>
When encoding in hybrid mode at low bitrate, we sometimes only have
enough bits to code a single CELT band (8 - 9.6 kHz). When that
happens, the second band (CELT band 18, from 9.6 to 12 kHz) cannot
use folding because it is wider than the amount already coded, and
falls back to LCG noise. Because it can also happen on transients
(e.g. stops), it can cause audible pre-echo.
To address the issue, we change the folding behavior so that it is
never forced to fall back to LCG due to the first band not containing
enough coefficients to fold onto the second band. This is achieved
by simply repeating part of the first band in the folding of the
second band. This changes the code in celt/bands.c around line 1237:
<span class="grey">Valin &amp; Vos Expires April 24, 2017 [Page 7]</span></pre>
<hr class='noprint' style='width: 96ex;' align='left'/><!--NewPage--><pre class='newpage'><a name="page-8" id="page-8" href="#page-8" class="invisible"> </a>
<span class="grey">Internet-Draft Opus Update October 2016</span>
b = 0;
}
- if (resynth &amp;&amp; M*eBands[i]-N &gt;= M*eBands[start] &amp;&amp; \
(update_lowband || lowband_offset==0))
+ if (resynth &amp;&amp; (M*eBands[i]-N &gt;= M*eBands[start] || \
i==start+1) &amp;&amp; (update_lowband || lowband_offset==0))
lowband_offset = i;
+ if (i == start+1)
+ {
+ int n1, n2;
+ int offset;
+ n1 = M*(eBands[start+1]-eBands[start]);
+ n2 = M*(eBands[start+2]-eBands[start+1]);
+ offset = M*eBands[start];
+ /* Duplicate enough of the first band folding data to \
be able to fold the second band.
+ Copies no data for CELT-only mode. */
+ OPUS_COPY(&amp;norm[offset+n1], &amp;norm[offset+2*n1 - n2], n2-n1);
+ if (C==2)
+ OPUS_COPY(&amp;norm2[offset+n1], &amp;norm2[offset+2*n1 - n2], \
n2-n1);
+ }
+
tf_change = tf_res[i];
if (i&gt;=m-&gt;effEBands)
{
as well as line 1260:
fold_start = lowband_offset;
while(M*eBands[--fold_start] &gt; effective_lowband);
fold_end = lowband_offset-1;
- while(M*eBands[++fold_end] &lt; effective_lowband+N);
+ while(++fold_end &lt; i &amp;&amp; M*eBands[fold_end] &lt; \
effective_lowband+N);
x_cm = y_cm = 0;
fold_i = fold_start; do {
x_cm |= collapse_masks[fold_i*C+0];
The fix does not impact compatibility, because the improvement does
not depend on the encoder doing anything special. There is also no
reasonable way for an encoder to use the original behavior to improve
quality over the proposed change.
<span class="grey">Valin &amp; Vos Expires April 24, 2017 [Page 8]</span></pre>
<hr class='noprint' style='width: 96ex;' align='left'/><!--NewPage--><pre class='newpage'><a name="page-9" id="page-9" href="#page-9" class="invisible"> </a>
<span class="grey">Internet-Draft Opus Update October 2016</span>
<span class="h2"><a class="selflink" name="section-10" href="#section-10">10</a>. Downmix to Mono</span>
The last issue is not strictly a bug, but it is an issue that has
been reported when downmixing an Opus decoded stream to mono, whether
this is done inside the decoder or as a post-processing step on the
stereo decoder output. Opus intensity stereo allows optionally
coding the two channels 180-degrees out of phase on a per-band basis.
This provides better stereo quality than forcing the two channels to
be in phase, but when the output is downmixed to mono, the energy in
the affected bands is cancelled sometimes resulting in audible
artefacts.
As a work-around for this issue, the decoder MAY choose not to apply
the 180-degree phase shift when the output is meant to be downmixed
(inside or outside of the decoder).
<span class="h2"><a class="selflink" name="section-11" href="#section-11">11</a>. New Test Vectors</span>
Changes in <a href="#section-9">Section 9</a> and <a href="#section-10">Section 10</a> have sufficient impact on the
testvectors to make them fail. For this reason, this document also
updates the Opus test vectors. The new test vectors now include two
decoded outputs for the same bitstream. The outputs with suffix 'm'
do not apply the CELT 180-degree phase shift as allowed in
<a href="#section-10">Section 10</a>, while the outputs without the suffix do. An
implementation is compliant as long as it passes either set of
vectors.
In addition, any Opus implementation that passes the original test
vectors from <a href="./rfc6716">RFC 6716</a> [<a href="./rfc6716" title="&quot;Definition of the Opus Audio Codec&quot;">RFC6716</a>] is still compliant with the Opus
specification. However, newer implementations SHOULD be based on the
new test vectors rather than the old ones.
The new test vectors are located at &lt;<a href="https://jmvalin.ca/misc_stuff/opus_newvectors.tar.gz">https://jmvalin.ca/misc_stuff/</a>
<a href="https://jmvalin.ca/misc_stuff/opus_newvectors.tar.gz">opus_newvectors.tar.gz</a>&gt;. (EDITOR: change to an ietf.org link when
ready)
<span class="h2"><a class="selflink" name="section-12" href="#section-12">12</a>. IANA Considerations</span>
This document makes no request of IANA.
Note to RFC Editor: this section may be removed on publication as an
RFC.
<span class="h2"><a class="selflink" name="section-13" href="#section-13">13</a>. Acknowledgements</span>
We would like to thank Juri Aedla for reporting the issue with the
parsing of the Opus padding.
<span class="grey">Valin &amp; Vos Expires April 24, 2017 [Page 9]</span></pre>
<hr class='noprint' style='width: 96ex;' align='left'/><!--NewPage--><pre class='newpage'><a name="page-10" id="page-10" href="#page-10" class="invisible"> </a>
<span class="grey">Internet-Draft Opus Update October 2016</span>
<span class="h2"><a class="selflink" name="section-14" href="#section-14">14</a>. References</span>
[<a name="ref-RFC2119" id="ref-RFC2119">RFC2119</a>] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", <a href="./bcp14">BCP 14</a>, <a href="./rfc2119">RFC 2119</a>,
DOI 10.17487/RFC2119, March 1997,
&lt;<a href="http://www.rfc-editor.org/info/rfc2119">http://www.rfc-editor.org/info/rfc2119</a>&gt;.
[<a name="ref-RFC6716" id="ref-RFC6716">RFC6716</a>] Valin, JM., Vos, K., and T. Terriberry, "Definition of the
Opus Audio Codec", <a href="./rfc6716">RFC 6716</a>, DOI 10.17487/RFC6716,
September 2012, &lt;<a href="http://www.rfc-editor.org/info/rfc6716">http://www.rfc-editor.org/info/rfc6716</a>&gt;.
Authors' Addresses
Jean-Marc Valin
Mozilla Corporation
331 E. Evelyn Avenue
Mountain View, CA 94041
USA
Phone: +1 650 903-0800
Email: jmvalin@jmvalin.ca
Koen Vos
vocTone
Email: koenvos74@gmail.com
Valin &amp; Vos Expires April 24, 2017 [Page 10]
</pre><br />
<span class="noprint"><small><small>Html markup produced by rfcmarkup 1.120, available from
<a href="https://tools.ietf.org/tools/rfcmarkup/">https://tools.ietf.org/tools/rfcmarkup/</a>
</small></small></span>
</div>
</body>
</html>
Reference in New Issue View Git Blame Copy Permalink