|
From: | Marcus Müller |
Subject: | Re: [Discuss-gnuradio] Audio Sink Buffer |
Date: | Sat, 5 Dec 2015 10:12:27 +0100 |
User-agent: | Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.3.0 |
Hi Kyle, the GNU Radio audio sink doesn't throw anything away. Technically, it just uses ALSA to send the audio samples to the specified (or default) sound card. It simply scales the input items, puts them into an internal buffer (d_buffer)[1], and instructs ALSA to write that buffer to the sound card [2]; nothing magical here. With PulseAudio, that soundcard usually is an emulated sound card, meant to make ALSA applications use pulse rather than directly sending audio to the sound card. So that'd be my focus when seeing such audio oddities. In fact, virtual consoles + pulse kind of ring my alarm bells: On most distributions (aka any I can think of), pulse gets started only when the user logs in to a graphical session (e.g. GNOME, KDE, Xfce, Mate...), and not when logging into a virtual console. I hence conclude that you're running such a graphical UI when you are starting a flow graph, which means Pulse should be there and ready to accept samples. The fact that switching out of and back into your session helps indicates that kicking Pulse out of and back into operation is the key; which probably means this is a pulse-internal issue, or even a sound card hardware issue. My approach here:
Hope we can get this out of the world, Best regards, [1] https://github.com/gnuradio/gnuradio/blob/master/gr-audio/lib/alsa/alsa_sink.cc#L350 [2] https://github.com/gnuradio/gnuradio/blob/master/gr-audio/lib/alsa/alsa_sink.cc#L494 On 03.12.2015 22:08, Kyle Chapman
wrote:
Hi, |
[Prev in Thread] | Current Thread | [Next in Thread] |