On Di, Feb 12 2013, Dmitri Paduchikh wrote: > Let me explain. I am blind and try to use these command-line utilities > to play a speech output. The controlling program waits for termination > of one player process and only then starts playing next portion of > speech. In this scenario such huge pauses (one or two seconds) are > rather painful. I am in the same situation here. I'm using Speech-dispatcher together with the Mbrola text-to-speech engine. Since Speech-dispatcher doesn't support Mbrola together with my german phoneme generator, I use a generic driver which uses some command line utilies which create the audio which then gets played by pactl. The playback starts with no noticeable delay, but after playback pactl or respectively paplay takes about two more seconds to finish. This adds a two second delay between i.e. two spoken sentences, which is quite annoying. I'm on Gentoo, my value for CONFIG_SND_HDA_PREALLOC_SIZE is 2048. I'm getting the same two seconds delay with my Logitech USB headset as well, which uses the snd-usb-audio kernel module, though. In my desktop PC I use a SB Audigy 4 which supports hardware mixing, so I can avoid PA for speech playback there, which is not a satisfying solution, though, as hardware mixing unfortunately isn't supported by newer sound cards any more. On my laptop with a Intel HDA soundcard and the mentioned USB headset, I do not have that choice. I would really appreciate any hint that helps me reducing the delay. thanks Henning
