Mingye Wang

escapefilenamelinuxlike could be better

Hi Rafael, Thanks for noticing this! I noticed that you are replying to the e-mail version: that one has some mistakes I later edited out. Sorry about that. I've played around with opus a bit more using the https://jmvalin.ca/opus/opus-1.3 sample files (see https://jmvalin.ca/opus/opus-1.3/speech_samples) and local opusenc, opusinfo, and ffmpeg. I found that 9 kbps sounds more clear than 8 regardless of VBR; inspection using a tool (you don't want to know about it, it's written in javascript -- it...

Oh wait, dream does have SET_BANDWIDTH already. Nothing to be done there -- just someone to test needed.

Found out that libopus has a demo program I can use to play with these settings. Another setting to investigate is frame size: SILK goes up to 60 ms frames to reduce frame overhead.

Oh, also the expected FAAC filename is: static const char* FaacLibNames[] = { "faac_drm.dll", "libfaac_drm.dll", "libfaac.dll", "faac.dll", nullptr }; And it looks like stuffing in AAC-LD is exactly what DREAM is doing with FAAC right now. I expect some tiny quality benefit from getting the ffmpeg AAC codec in instead. The basic AAC patents should have expired now, so all should be fine anyways. Real HE-AAC is still a patent firepit, not to mention xHE-AAC. On macOS you get a good encoder you are...

Man this sourceforge thing is terrible compared to GitHub. It's got no source code search! I initially assumed that opusSetupParam might be enforcing a better default, but it's only called from the decoding path. Changes needed to fix the above report EOPUSBandwidth and EOPUSSignalshould be given an additional "auto" member. DrmTransmitter.cpp should also be given auto as the default. Users should be instructed to prefer using EOPUSApplication for biasing the content detection, instead of overriding...

Man this sourceforge thing is terrible compared to GitHub. It's got no source code search! I initially assumed that opusSetupParam might be enforcing a better default, but it's only called from the decoding path. Changes needed to fix the above report EOPUSBandwidth and EOPUSSignalshould be given an additional "auto" member. DrmTransmitter.cpp should also be given auto as the default. Users should be instructed to prefer using EOPUSApplication for biasing the content detection, instead of overriding...

Man this sourceforge thing is terrible compared to GitHub. It's got no source code search! I initially assumed that opusSetupParam might be enforcing a better default, but it's only called from the decoding path. Changes needed to fix the above report EOPUSBandwidth and EOPUSSignalshould be given an additional "auto" member. DrmTransmitter.cpp should also be given auto as the default. Users should be instructed to prefer using EOPUSApplication for biasing the content detection, instead of overriding...