inference.sh

#!/bin/bash

# Standalone inference script for Saudi MSA Piper TTS
# Usage: ./inference.sh "Arabic text here" output.wav

set -e

# Configuration
MODEL_FILE="${MODEL_FILE:-saudi_msa_epoch455.onnx}"
CONFIG_FILE="${CONFIG_FILE:-saudi_msa_epoch455.onnx.json}"

# Colors for output
GREEN='\033[0;32m'
RED='\033[0;31m'
YELLOW='\033[1;33m'
NC='\033[0m' # No Color

# Function to print colored messages
print_success() {
    echo -e "${GREEN}✓${NC} $1"
}

print_error() {
    echo -e "${RED}✗${NC} $1"
}

print_info() {
    echo -e "${YELLOW}ℹ${NC} $1"
}

# Check if piper is installed
if ! command -v piper &> /dev/null; then
    print_error "piper command not found"
    echo ""
    echo "Please install piper-tts:"
    echo "  pip install piper-tts"
    echo ""
    echo "Or use the Python script:"
    echo "  python3 inference.py -t \"Arabic text\" -o output.wav"
    exit 1
fi

# Check if model file exists
if [ ! -f "$MODEL_FILE" ]; then
    print_error "Model file not found: $MODEL_FILE"
    echo ""
    echo "Please download the model from:"
    echo "  https://huggingface.co/ISTNetworks/saudi-msa-piper"
    exit 1
fi

# Get text and output file
if [ $# -eq 0 ]; then
    echo "Usage: $0 \"Arabic text\" [output.wav]"
    echo ""
    echo "Examples:"
    echo "  $0 \"مرحبا بك\" output.wav"
    echo "  echo \"مرحبا بك\" | $0 - output.wav"
    echo "  cat text.txt | $0 - output.wav"
    echo ""
    echo "Environment variables:"
    echo "  MODEL_FILE - Path to ONNX model (default: saudi_msa_epoch455.onnx)"
    echo "  CONFIG_FILE - Path to config file (default: saudi_msa_epoch455.onnx.json)"
    exit 1
fi

TEXT="$1"
OUTPUT="${2:-output.wav}"

# Read from stdin if text is "-"
if [ "$TEXT" = "-" ]; then
    print_info "Reading text from stdin..."
    TEXT=$(cat)
fi

# Check if text is empty
if [ -z "$TEXT" ]; then
    print_error "No text provided"
    exit 1
fi

# Synthesize
print_info "Model: $MODEL_FILE"
print_info "Text: $TEXT"
print_info "Output: $OUTPUT"
echo ""

echo "$TEXT" | piper --model "$MODEL_FILE" --output_file "$OUTPUT"

if [ $? -eq 0 ]; then
    print_success "Audio generated successfully: $OUTPUT"
    
    # Show file info if 'file' command is available
    if command -v file &> /dev/null; then
        file "$OUTPUT"
    fi
    
    # Show file size
    if command -v du &> /dev/null; then
        SIZE=$(du -h "$OUTPUT" | cut -f1)
        echo "File size: $SIZE"
    fi
else
    print_error "Failed to generate audio"
    exit 1
fi