Trail Wood Films

Trailwood Films & Media was working on Ravens in Yellowstone — a documentary about wildlife in the iconic U.S. national park. The story was lyrical, atmospheric, and rich with regional nuance. The narration needed to strike the perfect tone. But there was one catch:

They were short on time
They were budget-conscious
They had high output expectations

The ask was clear: Can we use AI to deliver professional narration faster — without losing story-engagement, accuracy, or control?

For many, this would have meant a tradeoff. But Dale took a different route by choosing VoiceArchive’s AI-hybrid voiceover solution.

A new way to deliver AI audio: Built on people, not scraped data

At VoiceArchive, we’ve developed something most providers don’t offer: a hybrid AI workflow, trained on real human voice artists who stay part of the process.

VoiceArchive’s AI-hybrid model blends synthetic voice technology with human collaboration and oversight — from training and consent to project feedback and corrections.

For this project, Dale used the voice model ENGUSM9725-AJ, a synthetic voice model trained with full consent by our human voice artist AJ. Unlike off-the-shelf AI tools, this approach came with a unique benefit: the artist behind the model was still part of the process.

“Our goal is to make AI voiceover a reliable tool — not a risk. That means combining synthetic speed with the care and accuracy of human guidance.”
— Anna Sticken, Head of Sales, VoiceArchive

The process: Designed for accuracy and ease

The entire project was completed using synthetic voice output — but with a human-designed workflow behind the scenes.

Script-based AI sample
Dale received a longer AI sample based on his actual script, helping him make a confident decision early.
Consent and project approval
The artist behind ENGUSM9725-AJ consented to the script, rate, and terms — a key part of our #fAIrVoices approach.
Full AI audio generation
The entire 5,000-word script was rendered using our internal AI production tools.
Client feedback and pronunciation fix
The word “coyote” was flagged for a more regionally accurate pronunciation. We handled the update quickly and easily.
Human fallback — not needed, but available
Had more complex issues come up, we could have escalated to the original human artist to re-record specific lines — a valuable option for project accuracy.

The result: A strong narrative – delivered fast

With no need for human correction, the project was completed quickly and smoothly — and with a tone that matched the film’s atmosphere.

“I was impressed by how natural the voice sounded — and how easy it was to give feedback and refine it. This workflow gave me both creative control and efficiency.”
— Dale Johnson, Director at Trailwood Films

For Dale, it was the rare case where AI didn’t mean compromise. And for us, it was another proof point that the right setup — not just the right tool — makes all the difference.

A note on ethics: How VoiceArchive approaches AI differently

This project also highlights the foundation of our synthetic voice strategy: respect for the artists behind the models.

We call it #fAIrVoices — and it shapes every decision we make:

Native language only: We don’t stretch voices beyond their original capability
Consent-based usage: Artists are involved in every project their voice is used for
Fair pay: Every project includes a clear and agreed-upon compensation model

📎 Read more about our AI policy here