Every interview I do ends up as an audio file on my phone. That's been true since I started freelancing five years ago — recorder app running, source talking, me nodding. The audio part was never the problem. What came after was.
I used to transcribe manually. Earbuds in, fingers on keyboard, pause-rewind-type on repeat. A 40-minute interview took me nearly two hours to get into a document I could actually work from. Once I started taking on more assignments — three or four stories per month, each requiring two or three interviews — that math stopped working.
I needed to transcribe audio to text free without paying per-minute fees that would eat into assignment rates or committing to a monthly subscription for a tool I might use unevenly. Here's the workflow I landed on, and why it's held up across a year of actual use.
The Problem With Most Free Transcription Options
Before I found what works, I cycled through the obvious alternatives.
Built-in dictation tools — the kind baked into Google Docs or Apple's accessibility features — require you to play the audio back through a speaker while dictation listens. That compounds background noise. Any room echo, street noise, or even the difference between your microphone and the source's voice degrades accuracy. I tried this twice and spent more time fixing errors than I would have spent transcribing manually.
Free tiers on paid transcription platforms are usually capped: 30 minutes per month, or three files, or some other limit that runs out before the week does. They're free in the sense that a hotel gym is free — only free if you're using the room.
Browser-based free tools with no account required tend to top out at short files. I regularly record interviews that run 35 to 55 minutes. Splitting them into chunks and stitching together separate transcripts is its own time cost.
What I Use Now: sipsip.ai
I started using sipsip.ai's transcriber after a colleague mentioned it in a Slack thread about podcast workflows. She was using it for a different reason — pulling quotes from recorded press briefings — but the use case translated directly.
The workflow is straightforward. I upload the M4A file from my phone's recorder app, or in cases where I've recorded a remote interview via Zoom, I upload the MP4 directly. No account setup, no credit card. The file processes and I get a timestamped transcript.
The first time I used it on a real assignment interview — a 44-minute conversation with a city planner about zoning reform — I was skeptical about accuracy. I read through the output against the audio on the parts I planned to quote. The word-for-word accuracy on clean speech was good enough that I could quote directly with only minor corrections for filler words. That's the bar I need: not perfect, but quotable with a review pass.
What I didn't expect was the timestamps. Every few seconds of transcript is anchored to a point in the audio. When my editor comes back to me asking me to verify a quote — which happens, as it should — I can jump directly to that moment in the file. That's saved me time I previously spent scrubbing audio to find the section I'd already transcribed.
My Actual Interview-to-Story Workflow
Here's how a typical assignment looks now:
During the interview: I use my phone's native recorder app. I don't overthink the setup — I keep the phone on the table between me and the source, roughly equidistant. Good enough audio comes from proximity, not equipment.
After the interview: I export the M4A and upload it directly to sipsip.ai. I don't wait until I'm back at my desk. If I'm between interviews or on the subway, I do the upload from my phone. By the time I sit down to write, the transcript is there.
Working the transcript: I paste it into a separate document and read through it with the audio open in another window. I flag three categories of things: direct quotes I'll use verbatim, paraphrases I'll attribute, and background context I won't attribute but need to understand the story. This part takes 20–30 minutes for a typical interview. Compare that to the two hours it took me to produce the same document manually.
On accuracy: I've noticed the tool handles clear, conversational speech well. Technical terminology and proper nouns — especially names — require a review pass regardless. A source named Ferreira might come through as Ferrera. A regulatory acronym might get mangled. None of that is surprising for any transcription tool. I treat the output as a first draft of the transcript, not a final record.
The Economics of Freelance Transcription
This is worth being direct about because it's why the free part matters.
Citation Capsule: A 2023 report by the Freelancers Union found that 74% of freelance journalists cited unpredictable tool costs as a top operational stressor. Transcription services that charge per audio minute typically run $0.10–$0.25 per minute, meaning a 45-minute interview costs $4.50–$11.25 per session — and a journalist doing 10 interviews per month accumulates $45–$112 in transcription costs before a story is even filed.
On a per-assignment rate of $300–$500 for a mid-tier outlet, that's a meaningful slice. Subscription tools with monthly fees solve the per-minute problem but introduce a fixed cost that doesn't scale down on slower months. When I'm on assignment, I use transcription heavily. When I'm between assignments, I might use it twice in three weeks. A flat monthly fee charges me the same either way.
Free transcription that actually works removes this calculation from my overhead entirely. That's not a small thing for someone billing project-to-project.
What I've Learned About Audio Quality
A year of using automated transcription has made me better at recording, which I didn't expect.
The clearest predictor of transcript quality is the gap between the speaker's voice and background noise — not the absolute volume of either. A quiet interview in a coffee shop with consistent ambient noise often transcribes better than a noisy one where a door keeps opening and closing unpredictably. Consistent background noise becomes part of the audio baseline. Sudden spikes confuse transcription models.
I've also started asking sources to pause before and after important points when I know I'll want clean quotes. It sounds like a small thing, and sources never notice I'm doing it, but it gives me clean sentence-start and sentence-end markers in the transcript.
Remote Interviews and Podcast Audio
About a third of my interviews happen over Zoom or a similar platform. I record locally using Zoom's built-in recording feature, which produces an MP4. sipsip.ai handles MP4 the same way it handles audio-only formats, which matters because I don't want to extract audio as a separate step before uploading.
I've also started using it for podcast interviews I want to quote. If I'm writing about a subject and a relevant expert appeared on a podcast, I can paste the YouTube link directly and get the transcript without downloading the episode first. That's become useful for fact-checking quotes I've seen floating around social media without a clear source — I find the original audio, transcribe it, and verify.
Citation Capsule: According to the Reuters Institute Digital News Report 2024, 62% of journalists surveyed use AI-assisted transcription tools at least occasionally, up from 31% in 2022. The shift is most pronounced among freelancers and smaller newsrooms without dedicated transcription staff or budget. Accuracy concerns remain the top reported barrier to full adoption, though satisfaction rates improve significantly when journalists report reviewing output before publication.
The One Workflow Change I'd Recommend
If you're a journalist, researcher, or anyone who regularly turns spoken-word audio into written documents, the single highest-value change you can make is removing manual transcription from your process entirely. Not because the accuracy is perfect — it isn't — but because the review-and-correct workflow is dramatically faster than transcribe-from-scratch, and the cognitive mode is different. Editing a draft is easier than producing one.
I've filed stories where sources read the final article and commented on how accurately I'd captured their words. That accuracy came from a combination of good transcript output and careful review — not from either one alone.
Free transcription that produces quotable output exists. For my workflow, it's sipsip.ai. The tool processes what I give it and returns something I can work with. That's all I needed.
FAQ
Is there a truly free way to transcribe audio to text?
Yes. sipsip.ai transcribes audio files and YouTube/podcast URLs at no cost, with no per-minute charges and no subscription required. You upload the file or paste a link, and the transcript is ready in roughly the same time as the audio's duration. There are no hidden fees for standard use.
How accurate is free audio transcription software?
Accuracy varies by tool and audio quality. For clear interview recordings with minimal background noise, modern free transcription tools using Whisper-based models typically hit 90–95% word accuracy. That's sufficient for quoting purposes when you review the output — which you should do regardless of the tool.
What is the best free tool to convert audio to text?
For journalists and researchers who need clean, quotable output without paying per minute, sipsip.ai is a strong option. It handles MP3, M4A, WAV, and MP4, and also works from YouTube links and podcast RSS feeds. The transcript includes speaker timestamps, which matters for multi-person interviews.
Can I transcribe a long interview for free?
Yes — tools like sipsip.ai do not cap transcript length the way some free-tier tools do. A 45-minute interview processes in full without file splitting. The limiting factor is usually your internet connection speed for the upload, not any service-side length restriction.
James Okafor is a freelance journalist and podcast reporter based in Lagos and London. He covers urban policy, infrastructure, and technology for regional and international outlets.
Ready to try it yourself? Upload your next interview audio at sipsip.ai — no account required, no per-minute charges.
Frequently asked questions
Freelance journalist James Okafor shares how he transcribes every source interview to text for free using sipsip.ai — no per-minute fees, no subscription, and output clean enough to quote directly in published pieces.
