Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Merge pull request #30 from espeak-ng/rel-1.3 release 1.3 Merge pull request #29 from DataTriny/oss-close-device Fix OSS device never getting closed Merge pull request #25 from barracuda156/darwin Fix headers for macOS Merge pull request #21 from brad0/oss_cleanup Cleanup header checks for OSS OpenBSD no longer uses OSS so remove the soundcard.h header check. FreeBSD / NetBSD / DragonFly / Linux all use sys/soundcard.h. Merge pull request #20 from codeofdusk/build-on-macos Check-in Mac fixes Merge pull request #18 from zment4/master Copy audio buffer and send for playback without blocking, fixes Windows audio choppyness Fix latency-related buffer sizing Turns out that eSpeak-NG (the main user of this lib) enforces a minimum buffer size of 60ms which is also the default size. This explains why smaller LATENCY values were inducing choppiness in the audio on some systems. Adjust the comment accordingly,. Also make sure computed buffer sizes don't land in the middle of a sample frame. Doing (samplerate * channels * LATENCY) / 1000 is wrong. Both ALSA and PulseAudio provide nice abstractions for buffer sizing so let's use them directly. In the ALSA case in particular, we want the period to be 60ms, not the whole buffer, so to interleave speech audio computation and audio playback. Fix audio choppiness on some systems Commit a41d46e816d2 ("Fix cancellation snappiness") made espeak unusable on the RaspberryPi due to extreme audio choppiness. This can sometimes be observed on some PC-type systems as well, albeit much less prominently. Relax the timing to the smallest value that makes it work again on the RaspberryPi. Issue #9: update notes about build environment
- Loading branch information