taco/microWakeWord-Trainer-Nvidia-Docker
1
0
Fork 0
mirror of https://github.com/TaterTotterson/microWakeWord-Trainer-Nvidia-Docker.git synced 2026-08-12 07:55:33 -06:00
Code Issues Packages Projects Releases Wiki Activity

Release NVIDIA WakeWord Trainer v27

Browse Source
This commit is contained in:
MasterPhooey
2026-08-10 07:06:44 -05:00
parent 68c4227cb7
commit 13f229e451
14 changed files with 572 additions and 262 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
cli/wake_word_sample_generator
View File

@@ -4,7 +4,7 @@ set -euo pipefail
PROGPATH="$(realpath "$0")"
PROGDIR="$(dirname "${PROGPATH}")"
KNOWN_ARGS=( samples batch-size data-dir language tts-mode tts-voice-count )
KNOWN_ARGS=( samples batch-size data-dir language english-accent tts-mode tts-voice-count )
# shellcheck source=/dev/null
source "${PROGDIR}/shell.functions"
WAKE_WORD="${POSITIONAL_ARGS[0]:-}"
@@ -17,12 +17,14 @@ fi
if [ "${HELP}" == "true" ] || [ -z "${WAKE_WORD}" ] ; then
cat <<EOF >&2
Usage: $0 [ --samples=<samples> ] [ --batch-size=<batch_size> ]
[ --language=<lang> ] [ --tts-mode=<modern|hybrid|piper> ]
[ --language=<lang> ] [ --english-accent=<accent> ]
[ --tts-mode=<modern|hybrid|piper> ]
[ --tts-voice-count=<voices> ] <wake_word>
--samples: Number of samples to generate. Default: ${DEFAULT_SAMPLES}
--batch-size: Generation batch size. Default: ${DEFAULT_BATCH_SIZE}
--language: TTS language code. Default: ${DEFAULT_LANGUAGE}
--english-accent: English accent emphasis. Default: ${DEFAULT_ENGLISH_ACCENT}
--tts-mode: modern, hybrid, or piper. Default: ${DEFAULT_TTS_MODE}
--tts-voice-count: Deprecated compatibility option; direct generation ignores it.
<wake_word> Required phrase to synthesize.
@@ -38,17 +40,31 @@ case "${TTS_MODE}" in
;;
esac
LANGUAGE="$(echo "${LANGUAGE}" | tr '[:upper:]' '[:lower:]')"
ENGLISH_ACCENT="$(echo "${ENGLISH_ACCENT}" | tr '[:upper:] -' '[:lower:]__')"
if [ "${LANGUAGE}" != "en" ]; then
ENGLISH_ACCENT="mixed"
fi
case "${ENGLISH_ACCENT}" in
mixed|australian|american|british|canadian|irish|scottish|new_zealand|indian|south_african) ;;
*)
echo "ERROR: unsupported --english-accent '${ENGLISH_ACCENT}'." >&2
exit 2
;;
esac
WORK_DIR="${DATA_DIR}/work"
SAMPLES_DIR="${WORK_DIR}/wake_word_samples"
mkdir -p "${WORK_DIR}"
START_TS=$EPOCHSECONDS
echo "===== Generating ${SAMPLES} wake-word samples (language=${LANGUAGE}, tts=${TTS_MODE}) ====="
echo "===== Generating ${SAMPLES} wake-word samples (language=${LANGUAGE}, accent=${ENGLISH_ACCENT}, tts=${TTS_MODE}) ====="
python3 "${PROGDIR}/tts_generate_samples.py" "${WAKE_WORD}" \
--samples="${SAMPLES}" \
--batch-size="${BATCH_SIZE}" \
--language="${LANGUAGE}" \
--english-accent="${ENGLISH_ACCENT}" \
--tts-mode="${TTS_MODE}" \
--voice-count="${TTS_VOICE_COUNT}" \
--data-dir="${DATA_DIR}" \