Hi OHF-Voice team,
First, thank you for stepping up to continue the development of Piper!
I am currently running a highly customized, low-latency text-to-speech architecture for real-time telemetry. To maintain extreme execution speeds and avoid micro-stutters on my application's main thread, my system hooks directly into the standalone C++ piper.exe binary via standard I/O (Memory Pipe Mode) to intercept raw PCM bytes.
It appears that recent releases (v1.4.2+) are exclusively distributed as Python wheel (.whl) packages. Injecting a Python interpreter wrapper around the TTS engine introduces unacceptable latency and CPU overhead for my strict real-time environment.
Could you please consider resuming the release of pre-compiled, standalone C++ binaries for Windows (similar to the piper_windows_amd64.zip from the original rhasspy releases)? This would allow users building low-latency environments to easily upgrade and benefit from your latest engine improvements without sacrificing raw execution speed.
Thank you!
Hi OHF-Voice team,
First, thank you for stepping up to continue the development of Piper!
I am currently running a highly customized, low-latency text-to-speech architecture for real-time telemetry. To maintain extreme execution speeds and avoid micro-stutters on my application's main thread, my system hooks directly into the standalone C++ piper.exe binary via standard I/O (Memory Pipe Mode) to intercept raw PCM bytes.
It appears that recent releases (v1.4.2+) are exclusively distributed as Python wheel (.whl) packages. Injecting a Python interpreter wrapper around the TTS engine introduces unacceptable latency and CPU overhead for my strict real-time environment.
Could you please consider resuming the release of pre-compiled, standalone C++ binaries for Windows (similar to the piper_windows_amd64.zip from the original rhasspy releases)? This would allow users building low-latency environments to easily upgrade and benefit from your latest engine improvements without sacrificing raw execution speed.
Thank you!