Synthetic Voices, the Uncanny Valley, and Voice Confrontation

How do we recognise and get past the uneasiness of hearing our own voice?

Perhaps it’s a while since you’ve heard a recording of your own voice. But for most people the first time hearing our voice played back is a strange, uncanny sensation.

“Is that really what I sound like,” is the most commonly uttered phrase. It can be jarring to the point of debilitating embarrassment for some, while others might simply avoid the process in the future where they can avoid it.

For others, it can be a source of amusement. For some it might even become a motivator to shift their speech patterns to be clearer or more deliberate if what they hear is a rather messy jumble that bears no relation to the audio track that plays their internal monologue.

Why does our own recorded voice sound so strange?

There’s science behind why we initially find our voice hard to recognise or immediately identify with. In fact, there is a term for it as well: voice confrontation. It essentially is loosely defined as when listeners do not like the sound of their recorded voice played back to them.

Part of our perception of our own, actively spoken voices come directly from the vibration of vocal cords producing the sounds so we’re not hearing with our ears alone as we do when hearing a recording of our voice. This discrepancy has been studied and highlights that when we speak, the sound of our voices is conducted via bone, tissue, and air with approximately equal intensity, and this maximises the sound we hear when we speak. And when we just hear a recording of ourselves via the air, it never sounds quite as ‘deep or as full’ as we hear when we speak live¹. This is voice confrontation, and one study found that 31% of individuals felt negatively about hearing their own voices; so if this is you, you aren’t alone. 

Phone voices, reading voices and everyday speech

In addition, people have different cadences and deliveries in particular circumstances. Many of us have worked with a colleague who has a “phone voice” that bears little relation to their more conversational tones. And, this too is completely normal, as people often change the way they speak in different environments and in relation to different stimuli. 

All of these factors make the first time we hear a recording of our own voice a strange and memorable experience. But for those around us, our speaking voice is a very strong part of our unique identity. It’s our calling card or signature sound, often our most memorable feature.

Our voice can be a source of comfort, reassurance, boundary setting, permission, love and profound wisdom.

Synthetic voices, the uncanny valley of voice confrontation, and what role personalised audiobooks can play

The so-called “uncanny valley” of the mismatch between our active and recorded voice has some applicability to our voice confrontation and sense of discomfort with synthetic and AI-generated voices, as well, particularly when they’re associated with photorealistic avatars and characters. 

This is a fascinating area when considering the development of personalised audiobooks, where people provide vocal samples which are then overlaid on treasured stories and narratives to create lasting creative works in unique, individual voices.

At a purely practical level, what is the experience like in creating a personalised audiobook in your own voice if when you hear the audiobook back you barely even recognise your voice? What if the experience of hearing your voice back is disconcerting or borderline uneasy?

In many ways this is why services developing personalised audiobooks have emphasised sharing as the key element.

We don’t often create personalised audiobooks in our own voices to listen back to them ourselves; we create them to share with others who find our voice remarkable, soothing, inspiring or any number of different emotions or feelings.

But perhaps there’s a practical hurdle to jump over when creating our personalised audiobooks. Maybe we need to brace for the uncanny valley of voice confrontation and steel ourselves against the uneasy feeling that hearing our record voice creates.

Personalised audio has a range of potential applications and benefits, from bedtime stories to memorialising treasured voices, to gift giving and community building. It’s a valuable new area of the publishing sector to explore. Hopefully the uncanny valley doesn’t present too wide an obstacle to traverse and we can start to build appreciative networks where recordings can bring comfort and reassurance to our friends and family who like nothing more than hearing the sound of our voice at the touch of a button.

1. The link in this para leads to a study that cites several other pieces of research covered here. For detailed info on each study, have a look.