Files
ki-story/speak.sh

26 lines
880 B
Bash
Executable File

#!/bin/sh
cd "$(dirname "$0")"
if test -f .env; then
set -a
. ./.env
set +a
fi
OUT="${1:-output.wav}"
TMP_TEXT=$(mktemp)
trap 'rm -f "$TMP_TEXT"' EXIT
cat > "$TMP_TEXT"
REF_AUDIO_B64=$(base64 -w 0 elena-referenz.wav)
ESCAPED_TEXT=$(awk '{gsub(/\\/, "\\\\"); gsub(/"/, "\\\""); gsub(/\t/, "\\t"); if (NR>1) printf "\\n"; printf "%s", $0}' "$TMP_TEXT")
printf '{"text":"%s","language":"German","ref_text":"Das Café Laube roch nach einer Mischung aus frisch gerösteten Bohnen, süßlichen Lilien und dem metallischen Unterton von Regen","trim_sample":true,"auto_denoise":false,"high_fidelity":false,"ref_audio_b64":"%s"}' "$ESCAPED_TEXT" "$REF_AUDIO_B64" | \
curl -X POST "https://vonkaiser-audiodojo.chutes.ai/tts/qwen3clone" \
-H "Authorization: Bearer $CHUTES_API_KEY" \
-H "Content-Type: application/json" \
--data-binary @- \
--output "$OUT"