Transcribe audio to text with AI: import limits, free tiers and real prices

Is transcribing a file you already recorded the same product as transcribing a live meeting?

No, and that is the confusion to clear up before you pay. All three tools in this guide are built around the online meeting: Otter joins Zoom, Teams and Meet, Fireflies joins calendar meetings on its own. Uploading a file you already have, an interview captured on a recorder, a workshop recorded on a phone, a podcast, is a side function with its own quota, and that quota is usually much tighter than live transcription.

Otter makes the gap obvious. Its free plan gives 300 transcription minutes per user per month, but only 3 audio or video file imports for the lifetime of the account. The two counters have nothing to do with each other. If what you need is live capture and a written recap, our meeting notes section covers that case separately.

Which import quotas are actually documented?

Only one of the three publishes a per plan upload number. Here is what the official pages we verified support.

Otter Business, at $30 monthly or $19.99 per user per month billed yearly, raises uploads to 6,000 file import minutes per user per month, with a 4 hour cap per meeting. Descript scales in media hours: 30 hours on Creator ($35 monthly, $24 billed yearly) and 40 hours on Business ($65 monthly, $50 billed yearly).

In our reading, Otter’s free plan is not a working tool for uploads: 3 files for the life of the account is a way to try the product, not a way to clear a batch of interviews. That is part of why its free plan scores 3 out of 5 in our rating, despite the 300 monthly minutes on the meeting side.

What audio formats do these tools accept?

We cannot tell you, and the gap matters. None of the official pages we verified for Otter, Fireflies and Descript publishes a list of supported extensions. Otter refers to audio or video files, Descript counts media hours, with no further detail in the sources we recorded.

Practical consequence: run your own file, in its original format, through a free plan before you subscribe. On Otter that burns 1 of your 3 lifetime imports, so start with the file that represents your real workload, not the shortest one you have.

Verbatim or summary: what does each plan give you?

The two outputs are not metered the same way. Word for word transcription draws on the minutes or media hours quota. Summaries, assistant questions and the features labeled as AI draw on a second counter, in credits or in queries.

Two traps deserve a flag. On Fireflies, the included AI credits are one time: 20 on Pro, 30 on Business, 50 on Enterprise, and they do not refill each month. The pricing page does not say so, the credits guide does. On Descript, media hours and AI credits are capped separately, so you can have hours left and no credits, or the reverse.

Do you still have to proofread the transcript?

Yes, and no vendor here publishes an accuracy rate that would let you budget that time. The price you pay says nothing about what you get on your own audio: a multi speaker meeting, a noisy room or dense industry vocabulary do not transcribe like one voice into a microphone.

What you can compare is the guardrails. Fireflies states 100+ languages and a multi language mode, several languages in one call, from the Business plan at $29 monthly or $19 per seat per month billed yearly. Otter transcribes 6 languages per its comparison table: English, Spanish, French, German, Japanese and Chinese. For a single English file both are covered, but an interview that switches language mid call points to Fireflies. Its features score is 4 out of 5, against 3 out of 5 for Otter.

Descript plays a different game: editing through the transcript, audio cleanup and captions. You fix the text and the media follows. Its features score, 4.5 out of 5, is the highest of the three.

Where do your audio files actually sit?

In the United States in two cases out of three, and somewhere unverified in the third. Otter stores data on AWS in the West region, with no published EU hosting option, and transfers from Europe rely on the Data Privacy Framework and standard contractual clauses. The vendor states a SOC 2 Type 2 report and AES-256 encryption, and also trains its own models on de-identified data. Its GDPR score is 2 out of 5.

Fireflies stores and processes in the US by default, on AWS and GCP. The private storage offered on Enterprise ($39 per seat per month, annual billing only) lets you keep meetings in your own AWS S3 or Google Cloud bucket, but processing stays in the US. The vendor states that customer data is not used to train AI models, with zero-day retention at its vendors, SOC 2 Type II and a Data Privacy Framework listing: GDPR score 3 out of 5.

For Descript, residency and contractual safeguards are not verified, which is why its GDPR score is 2.5 out of 5 and we assume no compliance. If data has to stay in Europe where you work, none of these three meets that in self serve. Another tool in our database, tl;dv, documents data centers in Europe, but it stays centered on meetings rather than file uploads.

One habit, whichever tool you pick: a recorded interview contains identifiable speech. Tell the people being recorded, and check the vendor’s retention terms before you upload sensitive files.

Which one fits your situation?

A few files a month, and a team already meeting online. Otter Pro, at $8.33 per user per month billed yearly, covers 10 monthly imports, 1,200 transcription minutes and a 90 minute cap per meeting. Past 10 files you have to move to Business.

A steady flow of interviews, in several languages. Fireflies Pro at $10 per seat per month billed yearly gives unlimited transcription and AI summaries, downloadable transcripts and recordings, and 100+ languages. Watch the storage, capped at 8,000 minutes per seat on that plan, and the free plan condition: unlimited free transcription only holds with auto-join for calendar meetings switched on, otherwise 3 transcriptions for accounts created since September 1, 2026.

Your audio file has to become something you publish. Descript Hobbyist, at $16 per person per month billed yearly, brings 10 media hours, 400 AI credits a month and watermark-free 1080p export. It is the only one of the three that edits the media from the corrected text.

In our reading, the deciding factor is neither the entry price nor the overall score, but the counter that blocks you first: the import count on Otter, storage on Fireflies, media hours on Descript. Run your real file through a free plan, time the proofreading it takes you, then decide.