Why we decided to Stop Using AI-Generated Voices

At Soundshape, we have explored AI voice technologies extensively. Over the past seven years, we tested a wide range of solutions, including TTS, Neural TTS, Speech to speech and voice-cloning pipeline.
We also trained our own datasets on voice’s team members to better understand how these technologies work.
We had many discussions with voice actors, voice directors, sound designers and voice designers.
We also worked with companies building AI based voice production datasets, and with clients explicitly requesting AI generated voices.
We delivered several projects involving AI voice technologies. And after three years of research and hands-on experience, we have decided to stop working with AI-generated voices.

This decision is driven by our core values : 

Quality :

Our projects require precise casting, accurate accents, and carefully directed emotional performances. We must be able to guide interpretation, intention and tone to match a specific audience, accent and context. Current AI voice systems do not meet the quality level required for the scale and impact of the campaigns and productions we work on.
‍
Trust :

Voice professionals do not trust AI voice generation, and we understand deeply why. Concerns include ethics, consent, fair remuneration, and transparent tracking of voice use. Building long-term relationships with our talents is a priority, and we are committed to preserving their trust and professional integrity.

Reliability :
‍
Our productions require very fast retakes.
While AI may appear faster on paper, it becomes unreliable as soon as it can’t pronounce specific phonemes or emotional intent that are not correct.
In practice, this forces a switch back to human recording, which introduces additional delays and production risks.

Data hosting and sovereignty :

Many of our clients, including European institutions, require strict and transparent data hosting policies. Data must remain hosted in Europe, and data sovereignty is a critical concern for our partners. Unless a voice generation system can be fully deployed and operated locally, this requirement cannot be properly satisfied.

Ethics :

Today, there is no clear and fair framework for voice rights, consent and long term compensation for performers.
The current ecosystem offers very limited transparency, and places voice talents in a structurally weak position.

For all these reasons, we have decided to stop offering AI generated voice content.

If you are looking for carefully selected professional voice talents, fast and structured production workflows, strict confidentiality, and the ability to operate within complex data hosting and compliance environments, then Soundshape is very likely the right partner for you.

Best,

François, Tanguy & Loïc,
Soundshape Management Team