Improve German TTS range handling
This commit is contained in:
@@ -15,3 +15,20 @@ cli_dir="${HERMES_HOME:-/opt/data}/.local/bin"
|
||||
mkdir -p "$cli_dir"
|
||||
ln -sfn /opt/hermes/.venv/bin/hermes "$cli_dir/hermes"
|
||||
ln -sfn /opt/hermes/.venv/bin/hermes /usr/local/bin/hermes
|
||||
|
||||
# Hermes streams each detected sentence to TTS separately. Its generic
|
||||
# sentence splitter treats the period in the German abbreviation "ca." as a
|
||||
# sentence ending, causing "ca." and the following measurement to be spoken
|
||||
# as two unrelated utterances. Expand the abbreviation before splitting.
|
||||
tts_target=/opt/hermes/tools/tts_streaming.py
|
||||
tts_old=' self.buf = _THINK_BLOCK_RE.sub("", self.buf + delta)'
|
||||
tts_new=' self.buf = _THINK_BLOCK_RE.sub("", self.buf + delta)
|
||||
self.buf = re.sub(r"\bca\.(?=\s)", "circa", self.buf, flags=re.IGNORECASE)'
|
||||
|
||||
if [ -f "$tts_target" ] && grep -Fq "$tts_old" "$tts_target"; then
|
||||
TTS_TARGET="$tts_target" TTS_OLD="$tts_old" TTS_NEW="$tts_new" python3 -c \
|
||||
'import os; from pathlib import Path; p = Path(os.environ["TTS_TARGET"]); s = p.read_text(); p.write_text(s.replace(os.environ["TTS_OLD"], os.environ["TTS_NEW"], 1))'
|
||||
echo "[tts-streaming-fix] expanded German ca. before sentence splitting"
|
||||
else
|
||||
echo "[tts-streaming-fix] upstream code already fixed or layout changed; no action"
|
||||
fi
|
||||
|
||||
Reference in New Issue
Block a user