Transcription for journalists

Quotes you can defend, timestamps you can prove

A misquote costs more than a correction; it costs the source's trust and your editor's, so Sonix is built so you never quote from memory or messy notes.

The math compounds on heavy reporting days. Three interviews before noon used to mean an evening of typing or a transcription invoice with a next-day turnaround. With Sonix all three are transcribed before lunch and searchable together, because the quote that makes your nut graf is sometimes in the interview you almost did not bother to check.

Every word is synced to the audio

Click any word in the transcript and the recording jumps to that exact moment. Before a contested quote runs, you verify it against the tape in seconds, not by scrubbing a 90-minute file.

Every voice gets a label

Sonix detects speakers automatically and labels them Speaker 1, Speaker 2, and so on. Rename them once and the names persist through the whole transcript. In a three-source story or a panel interview, you always know who said what. Here is how speaker diarization works under the hood.

Timecodes export with the text

Need to hand your editor or your legal review the quote plus the exact spot in the recording? Export with timestamps so anyone can check the audio themselves. For broadcast work, SRT export carries the timing straight into your edit.

This is what protects you from the misquote that memory produces. Handwritten notes compress what a source said into what you heard. A timestamped transcript keeps the exact wording, the hesitation before the answer, and the sentence that came right after, which is often the context a fairness read demands. When an editor asks whether she really said that, the answer is one click, played aloud.

Sonix delivers up to 99% accuracy on clear audio, and the browser-based editor plays the recording while you read, so cleaning up the remaining words takes a quick pass, not a second listen. For the full picture of how automated interview transcription works, see the interview transcription guide.

Security that takes your sources as seriously as you do

If a recording contains a sensitive source, where that file lives and who can touch it matters, so here is exactly what Sonix does with your audio.

• SOC 2 Type II certified. Controls are independently audited every year.
• Encrypted everywhere. TLS encryption in transit, AES-256 encryption at rest.
• Never used to train AI. None of your data processed through Sonix is used for training. Your interviews stay yours.
• Access on your terms. Two-factor authentication, role-based permissions, and password-protected sharing. Sonix employees cannot access your audio or transcripts unless you give express written consent to investigate an issue.
• Deletion that means deletion. Remove a file and both the audio and the transcript are wiped from the platform.

In practice, that means you can answer the question every careful editor eventually asks: where does the tape live, and who can hear it? With Sonix it lives encrypted on secure servers, visible only to the accounts you authorize, and gone for good when you delete it.

One honest caveat, because your sources deserve honesty

No transcription tool can promise anonymity. Protecting a confidential source still depends on your own practices, from how you record to what you name the file. What Sonix controls, it locks down, so the transcription step is never the weak link.

Full details, certifications, and compliance documents are on the security page.

Real conditions: field audio and foreign languages

Not every interview happens in a quiet studio, and we will not pretend otherwise.

What transcribes well

Clear speech recorded close to the microphone. A lav mic, a handheld recorder on the table, or a direct phone or video feed will land near the 99% accuracy mark.

What needs help

Heavy wind, crowd noise, speakers talking over each other, and recordings with a metallic ring from processed or compressed audio. For rough files, Sonix handles diverse accents and challenging conditions well, and noisy recordings can be run with the enhanced transcription option for better results. If your audio has that hollow, robotic edge, fix it before uploading: here is how to remove metallic sound from audio.

Reporting across borders

Sonix transcribes in 54+ languages, so the interview you recorded in Spanish, Arabic, or Mandarin comes back in the language it was spoken. Then translate the transcript for your newsroom, your editor, or a cross-border collaboration, and keep the original alongside it for verification.

Built for every interview you actually do

Phone calls, video calls, press conferences, and longform projects each produce a different recording, and the same timestamped transcript serves all four.

Phone interviews

Record the call with your usual setup, upload the file, and get a searchable phone call transcription in minutes. It works with any call recording your phone or recorder produces. Check your local consent laws before recording any call.

Video calls

Zoom, Meet, and Teams interviews upload as video and come back as speaker-labeled transcripts. See how Sonix transcribes meetings and video calls.

Press conferences

Multi-speaker detection separates the podium from the press pool, so a 45-minute briefing becomes an attributed Q&A you can quote from immediately. Archive search means you can check what an official said at last quarter's briefing, word for word, without replaying it.

Longform and documentary

When a project runs to dozens of interview hours, timestamped, speaker-labeled transcripts become your story map. Search every hour at once, compare how different subjects answered the same question, shape selects in text before you ever open the edit, and export SRT when you cut.

The best transcription software for journalists pairs speed with verifiability: fast turnaround, timestamps synced to audio, speaker labels, archive search, and security strong enough for sensitive recordings. The journalism transcription software guide compares the options in detail.

Pricing for freelancers and newsrooms alike

Sonix charges by the hour of audio you upload, prorated to the second, so a 40-minute interview bills as 40 minutes and never rounds up to the hour.

Filing occasionally

Pay As You Go is $10 per hour of audio with no subscription. One hour-long interview costs $10. A weekly interview habit runs about $40 a month, and you only pay when you upload.

Transcribing every week

Subscriptions drop the effective rate fast. Core is $25 per month with 5 hours included ($5 per hour). Advanced is $50 per month with 20 hours ($2.50 per hour). Pro is $80 per month with 40 hours ($2 per hour). Additional hours on any plan are $10 per hour, and annual billing costs 11 months, not twelve.

Compare that to human transcription services at $1 to $2 per minute, which puts a single hour-long interview at $60 to $120 with a turnaround measured in days, not minutes.

Working on a desk or a team

Extra seats are $25 per month each, and shared folders keep every reporter's transcripts searchable in one place. For collaboration, archive search across the whole team, and enterprise controls, see Sonix for newsrooms.

Full plan details are on the pricing page, or start with 30 free minutes and test it on your next interview.