lucataco · October 9, 2025 23:23
diff --git a/kokoro b/kokoro
 #!/bin/bash

 # This is local cli command that allows users to use kokoro on a Macbook Pro
 # Requires you to first run the kokoro docker container:
 docker run -p 8880:8880 ghcr.io/remsky/kokoro-fastapi-cpu:latest
 # Then save this file to /usr/local/bin
 # Finally you can test:
 kokoro "The quick brown fox jumped over the lazy dog"
 # Or even pipe from a stream like:
 llm "tell me a joke" | kokoro
 # See Examples below for more

 # Default values
 PORT=8880
 VOICE="af_sky"

 # Help message
 show_help() {
    cat << EOF
 Usage: kokoro [OPTIONS] [TEXT]
 Send text to local Kokoro TTS server and play the audio.
 Supports streaming - processes sentences as they arrive!

 Options:
    -p, --port PORT       Port number (default: 8880)
    -v, --voice VOICE     Voice to use (default: af_sky)
    -h, --help            Show this help message

 Examples:
    kokoro "Hello world"
    echo "Hello world" | kokoro
    llm "tell me a joke" | kokoro
    llm "explain quantum physics" | kokoro --voice af_bella
 EOF
    exit 0
 }

 # Parse arguments
 TEXT=""
 while [[ $# -gt 0 ]]; do
    case $1 in
        -p|--port)
            PORT="$2"
            shift 2
            ;;
        -v|--voice)
            VOICE="$2"
            shift 2
            ;;
        -h|--help)
            show_help
            ;;
        *)
            TEXT="$1"
            shift
            ;;
    esac
 done

 # Function to generate and queue audio
 generate_audio() {
    local text="$1"
    local output_file="$2"
    
    # Escape text for JSON
    local escaped_text=$(echo "$text" | python3 -c 'import json, sys; print(json.dumps(sys.stdin.read()))')
    
    # Generate audio
    curl -s -X POST "http://localhost:$PORT/v1/audio/speech" \
        -H "Content-Type: application/json" \
        -d "{\"input\":$escaped_text,\"voice\":\"$VOICE\",\"response_format\":\"wav\"}" \
        -o "$output_file" 2>/dev/null
    
    return $?
 }

 # Function to play audio file
 play_audio() {
    local file="$1"
    
    if [ ! -s "$file" ]; then
        return 1
    fi
    
    if command -v ffplay &> /dev/null; then
        ffplay -nodisp -autoexit -loglevel quiet "$file" 2>/dev/null
    elif command -v aplay &> /dev/null; then
        aplay -q "$file" 2>/dev/null
    elif command -v paplay &> /dev/null; then
        paplay "$file" 2>/dev/null
    elif command -v mpg123 &> /dev/null; then
        mpg123 -q "$file" 2>/dev/null
    else
        echo "Error: No audio player found" >&2
        return 1
    fi
 }

 # If text provided as argument, process it directly
 if [ -n "$TEXT" ]; then
    temp_file=$(mktemp "${TMPDIR:-/tmp}/kokoro.XXXXXX.wav")
    generate_audio "$TEXT" "$temp_file"
    play_audio "$temp_file"
    rm -f "$temp_file"
    exit 0
 fi

 # Streaming mode - process from stdin
 if [ ! -t 0 ]; then
    echo "🎤 Streaming mode active... (Press Ctrl+C to stop)" >&2
    
    # Directory for audio queue
    QUEUE_DIR=$(mktemp -d "${TMPDIR:-/tmp}/kokoro_queue.XXXXXX")
    
    # Cleanup function
    cleanup() {
        echo "" >&2
        echo "🛑 Stopping playback..." >&2
        
        # Kill all child processes
        pkill -P $$ 2>/dev/null
        
        # Kill player if it exists
        if [ -n "$PLAYER_PID" ]; then
            kill $PLAYER_PID 2>/dev/null
        fi
        
        # Remove queue directory
        rm -rf "$QUEUE_DIR" 2>/dev/null
        
        exit 0
    }
    
    # Set up trap for Ctrl+C
    trap cleanup SIGINT SIGTERM
    counter=0
    
    # Start player process
    (
        while true; do
            # Look for next audio file to play
            audio_file="$QUEUE_DIR/audio_$counter.wav"
            done_file="$QUEUE_DIR/done_$counter"
            
            # Wait for file to be ready
            while [ ! -f "$done_file" ]; do
                sleep 0.05
                # Check if we should exit
                if [ -f "$QUEUE_DIR/finished" ] && [ ! -f "$done_file" ]; then
                    exit 0
                fi
            done
            
            # Play the audio
            if [ -f "$audio_file" ]; then
                play_audio "$audio_file"
                rm -f "$audio_file" "$done_file"
            fi
            
            counter=$((counter + 1))
        done
    ) &
    PLAYER_PID=$!
    
    # Buffer for accumulating text
    buffer=""
    sentence_counter=0
    
    # Read and process input
    while IFS= read -r line || [ -n "$line" ]; do
        buffer="$buffer$line "
        
        # Extract complete sentences
        while [[ "$buffer" =~ ([^.!?]*[.!?][[:space:]]*) ]]; do
            sentence="${BASH_REMATCH[1]}"
            buffer="${buffer#$sentence}"
            sentence=$(echo "$sentence" | sed 's/^[[:space:]]*//;s/[[:space:]]*$//')
            
            if [ -n "$sentence" ]; then
                # Generate audio in background
                (
                    idx=$sentence_counter
                    audio_file="$QUEUE_DIR/audio_$idx.wav"
                    done_file="$QUEUE_DIR/done_$idx"
                    
                    generate_audio "$sentence" "$audio_file"
                    touch "$done_file"
                ) &
                
                sentence_counter=$((sentence_counter + 1))
            fi
        done
    done
    
    # Process remaining buffer
    if [ -n "$buffer" ]; then
        buffer=$(echo "$buffer" | sed 's/^[[:space:]]*//;s/[[:space:]]*$//')
        if [ -n "$buffer" ]; then
            audio_file="$QUEUE_DIR/audio_$sentence_counter.wav"
            done_file="$QUEUE_DIR/done_$sentence_counter"
            generate_audio "$buffer" "$audio_file"
            touch "$done_file"
        fi
    fi
    
    # Wait for all generation to complete
    wait
    
    # Signal player to finish
    touch "$QUEUE_DIR/finished"
    wait $PLAYER_PID 2>/dev/null
    
    # Cleanup
    rm -rf "$QUEUE_DIR"
    
    # Remove trap
    trap - SIGINT SIGTERM
 else
    echo "Error: No text provided"
    echo "Try 'kokoro --help' for more information"
    exit 1
 fi
	#!/bin/bash

	# This is local cli command that allows users to use kokoro on a Macbook Pro
	# Requires you to first run the kokoro docker container:
	docker run -p 8880:8880 ghcr.io/remsky/kokoro-fastapi-cpu:latest
	# Then save this file to /usr/local/bin
	# Finally you can test:
	kokoro "The quick brown fox jumped over the lazy dog"
	# Or even pipe from a stream like:
	llm "tell me a joke" \| kokoro
	# See Examples below for more

	# Default values
	PORT=8880
	VOICE="af_sky"

	# Help message
	show_help() {
	cat << EOF
	Usage: kokoro [OPTIONS] [TEXT]
	Send text to local Kokoro TTS server and play the audio.
	Supports streaming - processes sentences as they arrive!

	Options:
	-p, --port PORT Port number (default: 8880)
	-v, --voice VOICE Voice to use (default: af_sky)
	-h, --help Show this help message

	Examples:
	kokoro "Hello world"
	echo "Hello world" \| kokoro
	llm "tell me a joke" \| kokoro
	llm "explain quantum physics" \| kokoro --voice af_bella
	EOF
	exit 0
	}

	# Parse arguments
	TEXT=""
	while [[ $# -gt 0 ]]; do
	case $1 in
	-p\|--port)
	PORT="$2"
	shift 2
	;;
	-v\|--voice)
	VOICE="$2"
	shift 2
	;;
	-h\|--help)
	show_help
	;;
	*)
	TEXT="$1"
	shift
	;;
	esac
	done

	# Function to generate and queue audio
	generate_audio() {
	local text="$1"
	local output_file="$2"

	# Escape text for JSON
	local escaped_text=$(echo "$text" \| python3 -c 'import json, sys; print(json.dumps(sys.stdin.read()))')

	# Generate audio
	curl -s -X POST "http://localhost:$PORT/v1/audio/speech" \
	-H "Content-Type: application/json" \
	-d "{\"input\":$escaped_text,\"voice\":\"$VOICE\",\"response_format\":\"wav\"}" \
	-o "$output_file" 2>/dev/null

	return $?
	}

	# Function to play audio file
	play_audio() {
	local file="$1"

	if [ ! -s "$file" ]; then
	return 1
	fi

	if command -v ffplay &> /dev/null; then
	ffplay -nodisp -autoexit -loglevel quiet "$file" 2>/dev/null
	elif command -v aplay &> /dev/null; then
	aplay -q "$file" 2>/dev/null
	elif command -v paplay &> /dev/null; then
	paplay "$file" 2>/dev/null
	elif command -v mpg123 &> /dev/null; then
	mpg123 -q "$file" 2>/dev/null
	else
	echo "Error: No audio player found" >&2
	return 1
	fi
	}

	# If text provided as argument, process it directly
	if [ -n "$TEXT" ]; then
	temp_file=$(mktemp "${TMPDIR:-/tmp}/kokoro.XXXXXX.wav")
	generate_audio "$TEXT" "$temp_file"
	play_audio "$temp_file"
	rm -f "$temp_file"
	exit 0
	fi

	# Streaming mode - process from stdin
	if [ ! -t 0 ]; then
	echo "🎤 Streaming mode active... (Press Ctrl+C to stop)" >&2

	# Directory for audio queue
	QUEUE_DIR=$(mktemp -d "${TMPDIR:-/tmp}/kokoro_queue.XXXXXX")

	# Cleanup function
	cleanup() {
	echo "" >&2
	echo "🛑 Stopping playback..." >&2

	# Kill all child processes
	pkill -P $$ 2>/dev/null

	# Kill player if it exists
	if [ -n "$PLAYER_PID" ]; then
	kill $PLAYER_PID 2>/dev/null
	fi

	# Remove queue directory
	rm -rf "$QUEUE_DIR" 2>/dev/null

	exit 0
	}

	# Set up trap for Ctrl+C
	trap cleanup SIGINT SIGTERM
	counter=0

	# Start player process
	(
	while true; do
	# Look for next audio file to play
	audio_file="$QUEUE_DIR/audio_$counter.wav"
	done_file="$QUEUE_DIR/done_$counter"

	# Wait for file to be ready
	while [ ! -f "$done_file" ]; do
	sleep 0.05
	# Check if we should exit
	if [ -f "$QUEUE_DIR/finished" ] && [ ! -f "$done_file" ]; then
	exit 0
	fi
	done

	# Play the audio
	if [ -f "$audio_file" ]; then
	play_audio "$audio_file"
	rm -f "$audio_file" "$done_file"
	fi

	counter=$((counter + 1))
	done
	) &
	PLAYER_PID=$!

	# Buffer for accumulating text
	buffer=""
	sentence_counter=0

	# Read and process input
	while IFS= read -r line \|\| [ -n "$line" ]; do
	buffer="$buffer$line "

	# Extract complete sentences
	while [[ "$buffer" =~ ([^.!?][.!?][[:space:]]) ]]; do
	sentence="${BASH_REMATCH[1]}"
	buffer="${buffer#$sentence}"
	sentence=$(echo "$sentence" \| sed 's/^[[:space:]]//;s/[[:space:]]$//')

	if [ -n "$sentence" ]; then
	# Generate audio in background
	(
	idx=$sentence_counter
	audio_file="$QUEUE_DIR/audio_$idx.wav"
	done_file="$QUEUE_DIR/done_$idx"

	generate_audio "$sentence" "$audio_file"
	touch "$done_file"
	) &

	sentence_counter=$((sentence_counter + 1))
	fi
	done
	done

	# Process remaining buffer
	if [ -n "$buffer" ]; then
	buffer=$(echo "$buffer" \| sed 's/^[[:space:]]//;s/[[:space:]]$//')
	if [ -n "$buffer" ]; then
	audio_file="$QUEUE_DIR/audio_$sentence_counter.wav"
	done_file="$QUEUE_DIR/done_$sentence_counter"
	generate_audio "$buffer" "$audio_file"
	touch "$done_file"
	fi
	fi

	# Wait for all generation to complete
	wait

	# Signal player to finish
	touch "$QUEUE_DIR/finished"
	wait $PLAYER_PID 2>/dev/null

	# Cleanup
	rm -rf "$QUEUE_DIR"

	# Remove trap
	trap - SIGINT SIGTERM
	else
	echo "Error: No text provided"
	echo "Try 'kokoro --help' for more information"
	exit 1
	fi
No results found