<html><head><meta http-equiv="Content-Type" content="text/html charset=us-ascii"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; "><div>decodebin ! audioresample ! audioconvert ! plugin</div><div>plugin sink static caps: "<span style="color: rgb(0, 128, 0); ">audio/x-raw, format=(string)F32LE, layout=(string)interleaved"</span></div><div>I see your point. There must be some buffering plugin between the decoder and the rest - audioresample and audioconvert don't do that.</div><br><div><div>On Apr 9, 2013, at 8:44 PM, Wim Taymans <<a href="mailto:wim.taymans@gmail.com">wim.taymans@gmail.com</a>> wrote:</div><br class="Apple-interchange-newline"><blockquote type="cite">On 04/09/2013 07:23 PM, Tonu Jaansoo wrote:<br><blockquote type="cite">Hello!<br><br></blockquote><br>It depends on what your input is. If you have an mp3 decoder, it probably can only output one complete frame at a time.<br><br>For filesrc and other sources you could attempt to propose a bufferpool upstream, that should contain fixed sized buffers but again, it depends on upstream if this will actually be used.<br><br>What kind of input do you have?<br><br><blockquote type="cite">If i want to get exact multiple bytes of data from audioconvert plugin in GstBaseTransform audio filter plugin, what are my options?<br>The purpose of fixing input size is that currently each buffer that is coming into my plugin is in different size (10024 and 10032 alternating if i believe correctly). Because of that I would have to reallocate/change my internal stuff on each buffer - i do FFT on the data. I have tried to implement propose_allocation vmethod, but that method gives me correct size buffer on input (inside my transform vmethod), but the data is not filled using proposed size. How can i ask upstream element to give me nice chunk. Logically it should not be a problem for audioconvert - the size should just be divideable by channels*sizeof(typeofdata), right?<br><br>In an older and working hack I use not-in-place transform vmethod and two GstAdapters inside it - one for input and one for output. If there is way of fixing input buf size, I could avoid some memcpys.<br><br>Tonu<br>_______________________________________________<br>gstreamer-devel mailing list<br><a href="mailto:gstreamer-devel@lists.freedesktop.org">gstreamer-devel@lists.freedesktop.org</a><br>http://lists.freedesktop.org/mailman/listinfo/gstreamer-devel<br><br></blockquote><br>_______________________________________________<br>gstreamer-devel mailing list<br><a href="mailto:gstreamer-devel@lists.freedesktop.org">gstreamer-devel@lists.freedesktop.org</a><br>http://lists.freedesktop.org/mailman/listinfo/gstreamer-devel<br></blockquote></div><br></body></html>