Source: python-pymicro-wakeword
Maintainer: Home Assistant Team <team+homeassistant@tracker.debian.org>
Uploaders:
 Edward Betts <edward@4angle.com>,
Section: python
Build-Depends:
 debhelper-compat (= 13),
 dh-sequence-python3,
 pybuild-plugin-pyproject,
 python3-all,
 python3-setuptools,
Build-Depends-Indep:
 libtensorflow-lite2.14.1 <!nocheck>,
 python3-numpy <!nocheck>,
 python3-pymicro-features <!nocheck>,
 python3-pytest <!nocheck>,
Standards-Version: 4.7.4
Homepage: https://github.com/OHF-Voice/pymicro-wakeword
Vcs-Browser: https://salsa.debian.org/homeassistant-team/deps/python-pymicro-wakeword
Vcs-Git: https://salsa.debian.org/homeassistant-team/deps/python-pymicro-wakeword.git
Testsuite: autopkgtest-pkg-pybuild

Package: python3-pymicro-wakeword
Architecture: all
Depends:
 libtensorflow-lite2.14.1,
 ${misc:Depends},
 ${python3:Depends},
Description: wake word detection with microWakeWord models
 This library processes 16-bit mono audio at 16 kHz in small chunks, builds the
 feature frames expected by microWakeWord, and evaluates wake-word models to
 report detections or detection probabilities.
 .
 It can use bundled wake-word model files and associated metadata, including
 the built-in Okay Nabu model. Audio may be supplied by calling code as
 streaming frames, read from WAVE files, or provided on standard input from
 recording tools.
 .
 The command-line interface can run a selected model over one or more WAVE
 files or over a live raw audio stream. The library reports when the configured
 wake word is detected, and can also expose the probability value used for
 threshold-based detection.