Lesson 53 of 1570
Voice Cloning — Power and Ethics
ElevenLabs can clone a voice from 30 seconds of audio. That's useful for accessibility — and dangerous in the wrong hands. Here's how to use it well.
Lesson map
What this lesson covers
Learning path
The main moves in order
- 130 seconds of audio → a digital voice
- 2voice cloning
- 3ElevenLabs
- 4consent
Concept cluster
Terms to connect while reading
Section 1
30 seconds of audio → a digital voice
ElevenLabs v3 (released late 2025, 'Alpha' tier) is the current gold standard. Feed it 30-60 seconds of clean audio of a voice and it produces a synthetic voice that can say anything in that voice, in 30+ languages, with realistic emotion — whispering, laughing, singing. OpenAI's voice models and 11ai agents are close competitors.
Legitimate uses
- Accessibility — people losing their voice to ALS can bank a voice before it's gone (ElevenLabs partners with ALS Association).
- Audiobook narration — authors voice their own books in multiple languages.
- Podcast cleanup — fix flubbed words without re-recording.
- Character voices for indie game developers.
- Dubbing — translate your video into 30 languages in your own voice.
Abuses that are illegal and wrong
- Cloning a family member to run a 'grandparent scam' call.
- Cloning a CEO to authorize wire transfers (already caused multi-million-dollar losses in 2024-2025).
- Fake political robocalls (felonies in many US states since 2024).
- Cloning a person without consent for any purpose — harassment in 46 US states.
Using the ElevenLabs API
Python call to ElevenLabs v3 text-to-speech.
from elevenlabs.client import ElevenLabs
from elevenlabs import play
client = ElevenLabs(api_key="YOUR_KEY")
# Use an existing voice (your own, cloned with consent)
audio = client.text_to_speech.convert(
voice_id="your_voice_id_here",
model_id="eleven_v3", # v3 alpha, 2025
text="Hello, this is my voice reading a message I wrote.",
voice_settings={
"stability": 0.5,
"similarity_boost": 0.75,
"style": 0.3,
},
)
play(audio)Spotting voice clones
- 1Robotic breathing patterns — real breath is irregular.
- 2Perfect consistency — real voices vary volume, pace, and inflection.
- 3Delays or hesitations when asked unexpected questions (if it's a live call).
- 4Background absolutely silent — real calls have room noise.
ElevenLabs v3 special features
- Audio tags [laughs], [whispers], [sighs] — the model performs them.
- Emotional direction (happy, anxious, serious) via voice_settings.style.
- Multilingual — the same voice speaks 30+ languages natively.
- 11ai agents — voice + LLM + real-time conversation for customer service.
Compare the options
| Provider | Voice quality | Voice cloning speed | Consent policy |
|---|---|---|---|
| ElevenLabs v3 | Industry-leading; emotion + singing. | 30 sec → clone in minutes. | Voice verification challenge required. |
| OpenAI Advanced Voice | Very natural, conversational. | Limited custom voices; presets. | No user cloning in prod API. |
| Cartesia / Sonic | Very fast (real-time); good quality. | Quick clone. | Consent required. |
| Open-source (XTTS, StyleTTS2) | Decent; runs locally. | Depends on compute. | Self-policed. |
Key terms in this lesson
End-of-lesson quiz
Check what stuck
15 questions · Score saves to your progress.
Tutor
Curious about “Voice Cloning — Power and Ethics”?
Ask anything about this lesson. I’ll answer using just what you’re reading — short, friendly, grounded.
Progress saved locally in this browser. Sign in to sync across devices.
Related lessons
Keep going
Builders · 32 min
Video AI — Sora, Veo, Runway, Kling
Text-to-video became practical in 2025 and cinematic in 2026. Here's the state of the art and how to choose.
Builders · 26 min
Making Music with Suno and Udio
Type a prompt, get a full song — vocals, drums, mix, even in Portuguese. Here's how Suno v5, Udio, and ElevenMusic work — and what they can't yet do.
Builders · 30 min
Who Owns an AI Image?
US Copyright Office in 2026: works created purely by AI aren't copyrightable. Works with enough human creative control might be. Here's where the line sits right now.