server : implement "verbose_json" format with token details (#1781)

* examples/server: implement "verbose_json" format with token details.

This is intended to mirror the format of openai's Python
whisper.transcribe() return values.

* server: don't write WAV to a temporary file if not converting

* server: use std::lock_guard instead of manual lock/unlock
This commit is contained in:
Ryan Hitchman
2024-01-18 13:58:42 -07:00
committed by GitHub
parent fb466b3417
commit c0329acde8
3 changed files with 75 additions and 31 deletions

View File

@ -639,6 +639,12 @@ bool read_wav(const std::string & fname, std::vector<float>& pcmf32, std::vector
fprintf(stderr, "%s: read %zu bytes from stdin\n", __func__, wav_data.size());
}
else if (fname.size() > 256 || fname.size() > 40 && fname.substr(0, 4) == "RIFF" && fname.substr(8, 4) == "WAVE") {
if (drwav_init_memory(&wav, fname.c_str(), fname.size(), nullptr) == false) {
fprintf(stderr, "error: failed to open WAV file from fname buffer\n");
return false;
}
}
else if (drwav_init_file(&wav, fname.c_str(), nullptr) == false) {
fprintf(stderr, "error: failed to open '%s' as WAV file\n", fname.c_str());
return false;