#!/usr/bin/env bash
set -euo pipefail

TEXT="${1:-}"
OUT_WAV="${2:-}"

DEFAULT_VOICE="en_US-hal_6409-medium.onnx"
VOICE="${3:-$DEFAULT_VOICE}"
SPEED="${4:-1.0}"

PIPER_BIN="/home/sam/speech_piper/piper/piper"
MODEL_DIR="/home/sam/speech_piper/data"

if [[ -z "$TEXT" || -z "$OUT_WAV" ]]; then
  echo "Usage: $0 \"text\" /path/to/out.wav [voice.onnx] [speed]"
  exit 1
fi

TMP_WAV="/tmp/piper_doorbell_$$.wav"

echo "$TEXT" | \
  "$PIPER_BIN" --model "$MODEL_DIR/$VOICE" --output_file "$TMP_WAV" \
  --length-scale "$SPEED"

# Convert to ESP-friendly WAV: 16kHz mono 16-bit
sox "$TMP_WAV" -r 16000 -c 1 -b 16 -e signed-integer "$OUT_WAV"
rm -f "$TMP_WAV"

echo "Wrote: $OUT_WAV"