<?xml version="1.0" encoding="UTF-8"?>
<rss xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:taxo="http://purl.org/rss/1.0/modules/taxonomy/" version="2.0">
  <channel>
    <title>topic VAD for 8Khz? in Intel® Integrated Performance Primitives</title>
    <link>https://community.intel.com/t5/Intel-Integrated-Performance/VAD-for-8Khz/m-p/819705#M4590</link>
    <description>Hello Ed,&lt;BR /&gt;&lt;BR /&gt;I'm not familiar with the internals of the VAD sample, but if it has similar characteristics to the echo cancelation sample the following insights might be useful. &lt;BR /&gt;&lt;BR /&gt;Below is a quote from another thread regarding the application of 8 kHz data versus 16 kHz data for echo cancelation, from the engineer that implemented this code (he's on vacation right now). Perhaps the same notion can be applied to the VAD sample?&lt;BR /&gt;&lt;BR /&gt;Hope this helps,&lt;BR /&gt;&lt;BR /&gt;Paul&lt;BR /&gt;&lt;BR /&gt;&lt;BR /&gt;&lt;BLOCKQUOTE&gt;&lt;B&gt;Re: EC and NR&lt;/B&gt;&lt;BR /&gt;&lt;HR size="0" /&gt;&lt;DIV&gt;&lt;DIV id="quote_reply"&gt;&lt;BR /&gt;Hi&lt;BR /&gt;&lt;BR /&gt;IPP Echo canceller frame size is always equal to 160 bytes. Buffering is fully being done internally.EchoCancel function must be fed with data correspondent to 10 ms of 8 Khz and 5 ms of 16 Khz audio (which in case of linear PCMrepresented by 160 bytes both).So, for 20ms packet of 8KHz audio you shall to call it twice. &lt;BR /&gt;&lt;BR /&gt;EC works with linear PCM. I guess the noise problem you mentioned was caused bythat you fed a-law to EC.160 bytes of A-law represents 20 ms of 8KHz audio.After convertion you will get160 linear samples (320 bytes), i.e. two EC frames. You may do convertion by IPP functionippsAlawToLin_8u16s.&lt;BR /&gt;&lt;BR /&gt;Hope this help.&lt;BR /&gt;Vyacheslav&lt;/DIV&gt;&lt;/DIV&gt;&lt;/BLOCKQUOTE&gt;</description>
    <pubDate>Tue, 21 Sep 2010 22:33:48 GMT</pubDate>
    <dc:creator>PaulF_IntelCorp</dc:creator>
    <dc:date>2010-09-21T22:33:48Z</dc:date>
    <item>
      <title>VAD for 8Khz?</title>
      <link>https://community.intel.com/t5/Intel-Integrated-Performance/VAD-for-8Khz/m-p/819704#M4589</link>
      <description>Hello,&lt;DIV&gt;&lt;/DIV&gt;&lt;DIV&gt;Is there a VAD implementation for 8khz? The example is for 16khz and I was wondering if anyone has the filter parameters for 8khz sampling?&lt;/DIV&gt;&lt;DIV&gt;&lt;/DIV&gt;&lt;DIV&gt;Thanks,&lt;/DIV&gt;&lt;DIV&gt;Ed&lt;/DIV&gt;</description>
      <pubDate>Tue, 21 Sep 2010 18:45:36 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Integrated-Performance/VAD-for-8Khz/m-p/819704#M4589</guid>
      <dc:creator>Edward_Buchwalter</dc:creator>
      <dc:date>2010-09-21T18:45:36Z</dc:date>
    </item>
    <item>
      <title>VAD for 8Khz?</title>
      <link>https://community.intel.com/t5/Intel-Integrated-Performance/VAD-for-8Khz/m-p/819705#M4590</link>
      <description>Hello Ed,&lt;BR /&gt;&lt;BR /&gt;I'm not familiar with the internals of the VAD sample, but if it has similar characteristics to the echo cancelation sample the following insights might be useful. &lt;BR /&gt;&lt;BR /&gt;Below is a quote from another thread regarding the application of 8 kHz data versus 16 kHz data for echo cancelation, from the engineer that implemented this code (he's on vacation right now). Perhaps the same notion can be applied to the VAD sample?&lt;BR /&gt;&lt;BR /&gt;Hope this helps,&lt;BR /&gt;&lt;BR /&gt;Paul&lt;BR /&gt;&lt;BR /&gt;&lt;BR /&gt;&lt;BLOCKQUOTE&gt;&lt;B&gt;Re: EC and NR&lt;/B&gt;&lt;BR /&gt;&lt;HR size="0" /&gt;&lt;DIV&gt;&lt;DIV id="quote_reply"&gt;&lt;BR /&gt;Hi&lt;BR /&gt;&lt;BR /&gt;IPP Echo canceller frame size is always equal to 160 bytes. Buffering is fully being done internally.EchoCancel function must be fed with data correspondent to 10 ms of 8 Khz and 5 ms of 16 Khz audio (which in case of linear PCMrepresented by 160 bytes both).So, for 20ms packet of 8KHz audio you shall to call it twice. &lt;BR /&gt;&lt;BR /&gt;EC works with linear PCM. I guess the noise problem you mentioned was caused bythat you fed a-law to EC.160 bytes of A-law represents 20 ms of 8KHz audio.After convertion you will get160 linear samples (320 bytes), i.e. two EC frames. You may do convertion by IPP functionippsAlawToLin_8u16s.&lt;BR /&gt;&lt;BR /&gt;Hope this help.&lt;BR /&gt;Vyacheslav&lt;/DIV&gt;&lt;/DIV&gt;&lt;/BLOCKQUOTE&gt;</description>
      <pubDate>Tue, 21 Sep 2010 22:33:48 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Integrated-Performance/VAD-for-8Khz/m-p/819705#M4590</guid>
      <dc:creator>PaulF_IntelCorp</dc:creator>
      <dc:date>2010-09-21T22:33:48Z</dc:date>
    </item>
    <item>
      <title>VAD for 8Khz?</title>
      <link>https://community.intel.com/t5/Intel-Integrated-Performance/VAD-for-8Khz/m-p/819706#M4591</link>
      <description>Hi Edward,&lt;BR /&gt;As i know the VAD detection samples component processes input PCM file narrow band 16bit 8Khz or wide band 16bit 16Khz and creates output PCM file narrow band 16bit 8Khz or wide band 16bit 16khz respectively.&lt;BR /&gt;&lt;BR /&gt;Regards,&lt;BR /&gt;Navene Gv</description>
      <pubDate>Wed, 22 Sep 2010 03:08:29 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Integrated-Performance/VAD-for-8Khz/m-p/819706#M4591</guid>
      <dc:creator>Naveen_G_Intel</dc:creator>
      <dc:date>2010-09-22T03:08:29Z</dc:date>
    </item>
  </channel>
</rss>

