It's a classic scenario in any newsroom: a reporter comes back with a 20-minute interview. The slot in the show is three minutes. And the deadline is closing in. In the past, that meant headphones on, open the waveform, search manually, mark, cut, listen back, adjust. A process that could easily eat up an hour.

Today, there's a better way. AI can edit interviews automatically - precisely, with editorial intelligence, and in a fraction of the time. Here's how it works with Audrey Audio, the AI assistant built into the Radio Creator AI-Tools.

---

The Principle: Understand First, Then Cut

What sets Audrey apart from a traditional editing tool is simple: she understands the content. Automatic editing isn't based on waveform levels or manual markers - it's based on a transcript. Audrey reads the conversation, recognises context and meaning, and makes decisions like an experienced editor. Only in seconds.

The workflow runs in three steps:

1. Transcribe - Audrey converts the audio file into text, including precise timestamps for every segment.
2. Create an edit plan - Audrey analyses the transcript and determines which sections to keep and which to remove. You can give specific instructions or let Audrey suggest an approach.
3. Cut and listen - Audrey edits the audio automatically and delivers the result as an audio player. You can click directly on the text to jump to the corresponding point in the audio.

---

The Right Transcription Is the Key

For automatic editing to work precisely, the right transcription provider matters. For cutting, Audrey uses Mistral or AssemblyAI - both deliver the accurate segment timestamps that a reliable edit requires.

Already have a transcript from OpenAI? No problem: Audrey will automatically re-transcribe the file with the right provider on request.

A quick overview of the three available providers:

Provider	Strengths	Speaker Detection	Max. Length
Mistral	Most accurate, servers in Europe	✅ Yes	3 hours
AssemblyAI	Very long recordings, servers in Europe	✅ Yes	10 hours
OpenAI	Fastest processing	❌ No	50 minutes

For interviews with multiple speakers, Mistral or AssemblyAI are the go-to choice - they automatically detect who is speaking and display each speaker in a different colour in both the transcript and the waveform view.

---

Three Real-World Use Cases

1. Extracting Sound Bites

You've recorded a 20-minute interview with the mayor and only need the quotes about the new housing development? Audrey reads the transcript, identifies the relevant passages and cuts each sound bite as its own file. Every clip appears as a separate audio player - ready to preview and download.

Typical prompts:

• "Extract all statements from the mayor about the housing development as individual sound bites."
• "Find the three most compelling quotes from the interview and cut them as separate files."
• "Which statements work best as a sound bite for a 2-minute piece?"

2. Trimming an Interview to Broadcast Length

An interview runs 12 minutes, but the show only has room for 3? Audrey trims the piece to the required length - editorially sound, without cutting sentences in half. You can specify which topics must stay in, or let Audrey make a suggestion.

Typical prompts:

• "Trim the interview to around 3 minutes. Keep the statements about climate policy."
• "Remove all sections where the speaker repeats themselves."
• "Cut the piece down to the essentials - maximum 90 seconds."

3. Splitting an Interview Into Thematic Sections

A 45-minute expert interview needs to be split into three topic blocks for use across different programmes? Audrey analyses the content, proposes a structure and delivers three separate audio files - ready to use in different formats or podcast episodes.

---

Pro Tip: Enhance First, Then Transcribe

Anyone working with field recordings - vox pops, street interviews, phone-ins - knows the challenge: background noise, room reverb, traffic. Here, one extra step before transcription pays off.

The Audio Enhancer in the Radio Creator AI Tools removes background noise, improves speech clarity and optimises the overall sound. And that directly benefits the edit: a clean recording produces a more accurate transcript - and a more accurate transcript enables a more precise automatic cut.

The ideal workflow is: Upload audio → Audio Enhancer → Transcribe → Edit.

---

What AI Really Delivers in Interview Editing

Automatic interview editing with AI isn't magic - but it is a genuine productivity leap. What used to take an hour, Audrey handles in minutes. And she doesn't work blindly: she understands the content, respects context and never breaks a sentence mid-thought.

That makes AI editing especially valuable for:

• News teams that need sound bites for packages fast
• Podcast producers who want to structure long conversations
• Radio stations that need to bring interviews to broadcast length
• Agencies preparing interview material for multiple formats

---

Try It Yourself

Audrey Audio is part of the Radio Creator AI-Tools and runs directly in the browser - no installation, no prior audio production knowledge required. Simply upload the audio file, tell Audrey what you need, and listen to the result.

Give it a try - your next interview is already waiting.