Text-to-speech

Text-to-speech: give the robot a voice

Time: 10:15 AM to 10:55 AM

Your robot has a speaker and now you are going to make it talk. You will learn how the espeak text-to-speech engine works, build a program to explore different voices, and then combine speech with driving controls.

Test espeak from the terminal

Before writing any code, verify the speaker works:

espeak "Hello, I am PiCar X" 2>/dev/null

You should hear the robot speak. The 2>/dev/null suppresses audio system warnings — the speech still plays normally.

If you hear nothing, make sure the batteries are charged (above 7V) and run the audio setup:

cd ~/robot-hat && sudo bash i2samp.sh

Reboot and try again.

espeak parameters

espeak takes several flags that change how the speech sounds:

Flag	What it does	Example
`-s`	Speed in words per minute (default 175)	`-s 120` (slower)
`-p`	Pitch from 0–99 (default 50)	`-p 80` (higher)
`-v`	Voice / language	`-v en+f3` (female English)

Try these in the terminal:

espeak -s 80 "I am speaking very slowly" 2>/dev/null
espeak -s 250 "Now I am speaking fast" 2>/dev/null
espeak -v en+f3 "This is a female voice" 2>/dev/null
espeak -v de "Hallo, ich bin ein Roboter" 2>/dev/null

Program 1: TTS explorer

Build a menu-driven program that lets you try all the espeak options interactively.

nano ~/camp/day2/tts_demo.py

Step 1 — The speak function

This function wraps espeak so you can call it from Python. It pipes audio through PulseAudio with a direct-espeak fallback:

#!/usr/bin/env python3
import os
import sys
import time

def speak(text, speed=150, pitch=50, voice="en"):
    """Speak text using espeak."""
    safe = text.replace("'", "'\\''")
    ret = os.system(
        f"espeak --stdout -s {speed} -p {pitch} -v {voice} '{safe}' "
        f"2>/dev/null | paplay 2>/dev/null"
    )
    if ret != 0:
        os.system(
            f"espeak -s {speed} -p {pitch} -v {voice} '{safe}' 2>/dev/null"
        )

os.system() runs a shell command. The --stdout flag makes espeak output audio data instead of playing it directly, and paplay plays it through PulseAudio to the correct speaker. The safe variable escapes single quotes so phrases like “I’m a robot” do not break the command.

Step 2 — Demo functions

Each demo explores one parameter. Here is the speed demo:

def demo_speeds():
    print("\n  --- Different Speeds ---\n")
    text = "Speed test one two three"
    for speed in [80, 120, 175, 250]:
        label = {80: "slow", 120: "normal", 175: "fast", 250: "very fast"}
        print(f"  Speed {speed} ({label.get(speed, '')})...")
        speak(text, speed=speed)
        time.sleep(0.3)

The voice demo cycles through different languages and variants:

def demo_voices():
    print("\n  --- Different Voices ---\n")
    voices = [
        ("en",    "English (default)"),
        ("en+f3", "English (female)"),
        ("en+m7", "English (whisper)"),
        ("en-sc", "English (Scottish)"),
        ("de",    "German"),
        ("fr",    "French"),
        ("es",    "Spanish"),
    ]
    for voice_id, label in voices:
        print(f"  Voice: {label} ({voice_id})")
        speak("Hello, I am your robot.", voice=voice_id)
        time.sleep(0.3)

The custom message mode lets you type anything:

def demo_custom():
    print("\n  --- Custom Message ---")
    print("  Type a message and the robot will say it.")
    print("  Type 'done' to finish.\n")
    while True:
        msg = input("  Say: ").strip()
        if msg.lower() in ('done', 'quit', 'exit', 'q'):
            break
        if msg:
            speak(msg)

Wire the demos into a menu:

if __name__ == "__main__":
    print("=" * 50)
    print("  PiCar-X — Text-to-Speech Demo")
    print("=" * 50)

    speak("Ready", speed=175)

    print("""
  [1] Different speeds
  [2] Different voices
  [3] Type your own message
  [q] Quit
""")

    while True:
        choice = input("  Pick 1-3 or q: ").strip().lower()
        if choice == '1': demo_speeds()
        elif choice == '2': demo_voices()
        elif choice == '3': demo_custom()
        elif choice in ('q', 'quit'): break
        else: print("  Enter 1-3 or q.")

    print("\n  Bye!\n")

Run it

python3 ~/camp/day2/tts_demo.py

This script does not need sudo because it does not use motors or servos — it only plays audio.

Final code: tts_demo.py

Click to see the complete program

#!/usr/bin/env python3
import os
import sys
import time


def speak(text, speed=150, pitch=50, voice="en"):
    safe = text.replace("'", "'\\''")
    ret = os.system(f"espeak --stdout -s {speed} -p {pitch} -v {voice} '{safe}' 2>/dev/null | paplay 2>/dev/null")
    if ret != 0:
        ret = os.system(f"espeak -s {speed} -p {pitch} -v {voice} '{safe}' 2>/dev/null")
    return ret == 0


def demo_basic():
    print("\n  --- 1. Basic Speech ---\n")
    phrases = [
        "Hello! I am PiCar X.",
        "I can talk using text to speech.",
        "Pretty cool, right?",
    ]
    for phrase in phrases:
        print(f"  Saying: \"{phrase}\"")
        speak(phrase)
        time.sleep(0.5)


def demo_speeds():
    print("\n  --- 2. Different Speeds ---\n")
    text = "Speed test one two three"
    for speed in [80, 120, 175, 250]:
        label = {80: "slow", 120: "normal", 175: "fast", 250: "very fast"}
        print(f"  Speed {speed} ({label.get(speed, '')})...")
        speak(text, speed=speed)
        time.sleep(0.3)


def demo_voices():
    print("\n  --- 3. Different Voices ---\n")
    voices = [
        ("en",    "English (default)"),
        ("en+f3", "English (female)"),
        ("en+m7", "English (whisper)"),
        ("en-sc", "English (Scottish)"),
        ("de",    "German"),
        ("fr",    "French"),
        ("es",    "Spanish"),
    ]
    for voice_id, label in voices:
        print(f"  Voice: {label} ({voice_id})")
        speak("Hello, I am your robot.", voice=voice_id)
        time.sleep(0.3)


def demo_pitch():
    print("\n  --- 4. Different Pitches ---\n")
    for pitch in [10, 30, 50, 70, 99]:
        label = "low" if pitch < 30 else "mid" if pitch < 60 else "high"
        print(f"  Pitch {pitch} ({label})...")
        speak("Testing pitch", pitch=pitch)
        time.sleep(0.3)


def demo_custom():
    print("\n  --- 5. Custom Message ---")
    print("  Type a message and the robot will say it.")
    print("  Type 'done' to finish.\n")
    while True:
        msg = input("  Say: ").strip()
        if msg.lower() in ('done', 'quit', 'exit', 'q'):
            break
        if msg:
            speak(msg)


if __name__ == "__main__":
    print("=" * 50)
    print("  PiCar-X — Text-to-Speech Demo")
    print("=" * 50)

    if not speak("Ready", speed=175):
        sys.exit(1)

    print("""
  [1] Basic speech    [2] Different speeds
  [3] Different voices [4] Different pitches
  [5] Custom message   [a] Run all  [q] Quit
""")

    while True:
        choice = input("  Pick 1-5, a, or q: ").strip().lower()
        if choice == '1': demo_basic()
        elif choice == '2': demo_speeds()
        elif choice == '3': demo_voices()
        elif choice == '4': demo_pitch()
        elif choice == '5': demo_custom()
        elif choice == 'a':
            demo_basic(); demo_speeds(); demo_voices()
            demo_pitch(); demo_custom()
        elif choice in ('q', 'quit'): break
        else: print("  Enter 1-5, a, or q.")

    print("\n  Bye!\n")

Program 2: talking driver

Now combine speech with movement. The robot announces each action before performing it. Speech runs in a background thread so the robot keeps driving while it talks.

nano ~/camp/day2/talking_driver.py

Step 1 — Background speech

The key difference from tts_demo.py: speech must not block driving. Use threading to run espeak in the background:

#!/usr/bin/env python3
import os
import sys
import time
import threading

from picarx import Picarx

def getch():
    import tty, termios
    fd = sys.stdin.fileno()
    old = termios.tcgetattr(fd)
    try:
        tty.setraw(fd)
        ch = sys.stdin.read(1)
    finally:
        termios.tcsetattr(fd, termios.TCSADRAIN, old)
    return ch

def speak(text, speed=175):
    """Speak in a background thread so driving does not pause."""
    def _say():
        safe = text.replace("'", "'\\''")
        os.system(f"espeak -s {speed} '{safe}' 2>/dev/null")
    threading.Thread(target=_say, daemon=True).start()

threading.Thread(target=_say, daemon=True).start() launches _say in a separate thread. The main loop immediately continues to the next keypress without waiting for espeak to finish. daemon=True means the thread automatically stops when the program exits.

Step 2 — The control loop

Same WASD pattern as the keyboard driver, but each action calls speak() first:

px = Picarx()
speed = 35
steer_angle = 0
steer_step = 10
CLEAR = "\033[2K\r"

speak("Ready to drive")
time.sleep(1)

print("  Controls: W=fwd S=back A=left D=right SPACE=stop Q=quit\n")

try:
    while True:
        ch = getch()

        if ch in ('q', 'Q', '\x03'):
            speak("Goodbye")
            time.sleep(1)
            break
        elif ch in ('w', 'W'):
            speak("Forward")
            px.forward(speed)
            sys.stdout.write(f"{CLEAR}  >> Forward at {speed}%")
            sys.stdout.flush()
        elif ch in ('s', 'S'):
            speak("Backward")
            px.backward(speed)
            sys.stdout.write(f"{CLEAR}  >> Backward at {speed}%")
            sys.stdout.flush()
        elif ch in ('a', 'A'):
            steer_angle = max(steer_angle - steer_step, -35)
            speak("Turning left")
            px.set_dir_servo_angle(steer_angle)
            sys.stdout.write(f"{CLEAR}  >> Steer {steer_angle:+d}°")
            sys.stdout.flush()
        elif ch in ('d', 'D'):
            steer_angle = min(steer_angle + steer_step, 35)
            speak("Turning right")
            px.set_dir_servo_angle(steer_angle)
            sys.stdout.write(f"{CLEAR}  >> Steer {steer_angle:+d}°")
            sys.stdout.flush()
        elif ch == ' ':
            speak("Stopping")
            px.stop()
            steer_angle = 0
            px.set_dir_servo_angle(0)
            sys.stdout.write(f"{CLEAR}  >> Stopped")
            sys.stdout.flush()

except KeyboardInterrupt:
    pass
finally:
    px.stop()
    px.set_dir_servo_angle(0)
    print(f"\n\n  Stopped. Bye!\n")

Run it

sudo python3 ~/camp/day2/talking_driver.py

Drive the robot around and listen — it says “Forward”, “Turning left”, “Stopping”, etc. as you press keys.

Final code: talking_driver.py

Click to see the complete program

#!/usr/bin/env python3
import os
import sys
import time
import threading

from picarx import Picarx


def getch():
    import tty, termios
    fd = sys.stdin.fileno()
    old = termios.tcgetattr(fd)
    try:
        tty.setraw(fd)
        ch = sys.stdin.read(1)
    finally:
        termios.tcsetattr(fd, termios.TCSADRAIN, old)
    return ch


def speak(text, speed=175):
    def _say():
        safe = text.replace("'", "'\\''")
        os.system(f"espeak -s {speed} '{safe}' 2>/dev/null")
    threading.Thread(target=_say, daemon=True).start()


px = Picarx()
speed = 35
steer_angle = 0
steer_step = 10
CLEAR = "\033[2K\r"

print("=" * 50)
print("  PiCar-X — Talking Driver")
print("=" * 50)
print("  Controls: W=fwd S=back A=left D=right SPACE=stop Q=quit\n")

speak("Ready to drive")
time.sleep(1)

try:
    while True:
        ch = getch()

        if ch in ('q', 'Q', '\x03'):
            speak("Goodbye")
            time.sleep(1)
            break
        elif ch in ('w', 'W'):
            speak("Forward")
            px.forward(speed)
            sys.stdout.write(f"{CLEAR}  >> Forward at {speed}%")
            sys.stdout.flush()
        elif ch in ('s', 'S'):
            speak("Backward")
            px.backward(speed)
            sys.stdout.write(f"{CLEAR}  >> Backward at {speed}%")
            sys.stdout.flush()
        elif ch in ('a', 'A'):
            steer_angle = max(steer_angle - steer_step, -35)
            speak("Turning left")
            px.set_dir_servo_angle(steer_angle)
            sys.stdout.write(f"{CLEAR}  >> Steer {steer_angle:+d}°")
            sys.stdout.flush()
        elif ch in ('d', 'D'):
            steer_angle = min(steer_angle + steer_step, 35)
            speak("Turning right")
            px.set_dir_servo_angle(steer_angle)
            sys.stdout.write(f"{CLEAR}  >> Steer {steer_angle:+d}°")
            sys.stdout.flush()
        elif ch == ' ':
            speak("Stopping")
            px.stop()
            steer_angle = 0
            px.set_dir_servo_angle(0)
            sys.stdout.write(f"{CLEAR}  >> Stopped")
            sys.stdout.flush()

except KeyboardInterrupt:
    pass
finally:
    px.stop()
    px.set_dir_servo_angle(0)
    print(f"\n\n  Stopped. Bye!\n")

Challenge: customize the talking driver

Custom startup message

Change the speak("Ready to drive") line to your own greeting. The robot will say it every time the script starts.

Add speed announcements

Add elif blocks for + and - keys that announce the new speed. For example: speak(f"Speed {speed} percent").

Custom voice

Change the voice in the speak() function. Try espeak -v en+f3 for a female voice, or -v en+m7 for a whisper.

After this section, take a 10-minute break (10:55 AM to 11:05 AM).

Welcome

Class Recordings

Day 1: Setup and Calibration

Day 2: Code & Computer Vision

Day 3: GenAI and Cloud LLMs

Day 4: Vision AI

Day 5: AI Ethics & Final Project

Text-to-speech: give the robot a voice

Test espeak from the terminal

espeak parameters

Program 1: TTS explorer

Step 1 — The speak function

Step 2 — Demo functions

Step 3 — Menu

Run it

Final code: tts_demo.py

Program 2: talking driver

Step 1 — Background speech

Step 2 — The control loop

Run it

Final code: talking_driver.py

Challenge: customize the talking driver

​Text-to-speech: give the robot a voice

​Test espeak from the terminal

​espeak parameters

​Program 1: TTS explorer

​Step 1 — The speak function

​Step 2 — Demo functions

​Step 3 — Menu

​Run it

​Final code: tts_demo.py

​Program 2: talking driver

​Step 1 — Background speech

​Step 2 — The control loop

​Run it

​Final code: talking_driver.py

​Challenge: customize the talking driver

Text-to-speech: give the robot a voice

Test espeak from the terminal

espeak parameters

Program 1: TTS explorer

Step 1 — The speak function

Step 2 — Demo functions

Step 3 — Menu

Run it

Final code: tts_demo.py

Program 2: talking driver

Step 1 — Background speech

Step 2 — The control loop

Run it

Final code: talking_driver.py

Challenge: customize the talking driver