In an era where artificial intelligence is becoming increasingly sophisticated, distinguishing between real and AI-generated voices can be challenging. Fake AI voices are often used in scams, misinformation campaigns, and other malicious activities. Understanding how to spot these synthetic voices can help you protect yourself from potential fraud. Daniel Harding, a cyber security expert from MaxContact explains what you need to know about identifying fake AI voices and the red flags to watch out for.
The Rise of Fake AI Voices
With advancements in AI technology, creating realistic-sounding synthetic voices has become easier. These AI-generated voices can mimic human speech patterns and even replicate the voice of specific individuals. While this technology has legitimate applications, such as in virtual assistants and accessibility tools, it can also be misused.
Why Spotting Fake AI Voices is Important
Fake AI voices can be used in various deceptive ways, including:
Scams: Impersonating individuals to trick people into sharing personal information or transferring money.
Misinformation: Spreading false information by mimicking trusted voices.
Identity Theft: Replicating someone’s voice to gain unauthorised access to accounts and services.
Key Indicators of Fake AI Voices
Here are some tips and techniques to help you spot fake AI voices:
AI-generated voices might struggle with the natural flow of conversation. Listen for:
- Monotone Delivery: Lack of natural intonation and emphasis.
- Awkward Pauses: Unnatural breaks in speech that disrupt the flow of sentences.
- Over-Enunciation: Pronouncing every syllable too clearly, which can sound robotic.
Inconsistent Pronunciation
Fake AI voices often mispronounce words, especially complex ones. Some keywords that can be particularly telling include:
Anemone – Often pronounced as “an-e-mo-nee” instead of “uh-NEM-uh-nee.”
Quinoa – Mispronounced as “kwee-no-uh” rather than “KEEN-wah.”
Worcestershire – Commonly butchered as “wor-ces-ter-shy-er” instead of “WUSS-ter-sheer.”
Isthmus – Mispronounced as “isth-mus” instead of “is-muss.”
Hermione – AI might pronounce it as “her-mee-own” rather than “her-MY-oh-nee.”
Mispronounced Names – or Luck of the Irish?
Certain names are frequently mispronounced by AI, which can be a strong indicator of synthetic voices. Watch for mispronunciations of names like:
Siobhan – Mispronounced as “See-oh-ban” instead of “Shiv-on.”
Niamh – Mispronounced as “Nee-am” instead of “Neev.”
Aoife – Mispronounced as “Ay-oh-fee” instead of “Ee-fa.”
Saoirse – Mispronounced as “Say-or-suh” instead of “Seer-sha.”
Eoghan – Mispronounced as “Ee-oh-gan” instead of “Owen.”
Tadhg – Mispronounced as “Tad-hig” instead of “Tige.”
Caoimhe – Mispronounced as “Kay-oh-mee” instead of “Kee-va.”
Róisín – Mispronounced as “Roy-sin” instead of “Roh-sheen.”
Maëlle – Mispronounced as “May-ell” instead of “My-el.”
Lachlan – Mispronounced as “Latch-lan” instead of “Lock-lin.”
Lack of Emotional Nuance
Human voices convey emotions naturally, while AI voices may lack this depth. Signs include:
Flat Emotional Tone: Difficulty conveying excitement, sadness, or anger.
Inconsistent Emotional Context: Sudden shifts in emotion that don’t match the content of the conversation.
Technical Glitches
AI voices can sometimes glitch or produce odd sounds that are not typical of human speech
Distorted Sounds: Artefacts or digital noise within the voice.
Repetition: Repeating certain phrases or words in an unnatural manner.
Protecting Yourself from Fake AI Voices
Being aware of the potential for fake AI voices is the first step in protecting yourself. Here are some additional strategies:
Verify the Source
If you receive a suspicious call or message, verify the identity of the speaker through another trusted communication channel. For example, if your bank calls you, hang up and call the bank back using the official number from their website.
Use Authentication Tools
Many services offer voice authentication tools that can help confirm the identity of the speaker. Utilise these features when available to add an extra layer of security.
Cyber expert Daniel Harding from MaxContact AUS says:
While AI voice technology offers many benefits, it also poses significant risks when used maliciously. By learning to identify the signs of fake AI voices, you can protect yourself from scams and misinformation. Pay attention to speech patterns, pronunciation inconsistencies, emotional nuances, and technical glitches to spot synthetic voices. Stay vigilant, verify sources, and use available authentication tools to ensure your interactions are safe and secure.
