A podcast archive is hours of mostly unedited speech, and most transcription services charge for every minute of it. templobe runs an English speech model on your Mac instead: queue a season's worth of episodes, export srt and vtt for captions, and keep guest audio off anyone else's servers.
mp3, wav, m4a. Up to 50 files per batch. No upload. English only in this beta.
| What matters | templobe | Typical cloud service |
|---|---|---|
| Cost of a 60-minute episode | Free for beta users | One hour of billable minutes |
| Cost of the back catalogue | Free for beta users | Scales with every archived minute |
| Caption formats | srt and vtt | Usually srt and vtt |
| Show notes source | txt or md export | Copy from a web editor |
| Where guest audio goes | Stays on your Mac | Uploaded to a vendor |
| Works offline | Yes, after one sign-in | No |
| Batch the backlog | Up to 50 files per batch | Often one upload at a time |
| Speaker labels | Not in this beta | Offered by some providers |
This compares the common cloud model, not one vendor. Confirm guest and co-host consent terms before sending episode audio to any third party.
Select the exported mp3, wav or m4a files. Up to 50 per batch, 2 GB per file and 5 GB total — enough for several long episodes or a whole short-form season.
The model installed on your Mac transcribes each file. No upload, no queue on someone else's server, and the machine can stay offline.
Use srt or vtt for captions and video platforms, and txt or md for show notes, clips and search.
Yes. Episodes are handled as long audio and split into segments internally. In our benchmark the long-audio set scored 1.26% word error rate, but that figure comes from only two clips — treat it as indicative rather than a guarantee for hour-long recordings.
SubRip (.srt) and WebVTT (.vtt), plus plain text and Markdown for show notes and scripts.
No. There are no per-minute credits. A 60-minute episode costs the same as a two-minute clip during the beta: your Mac's processing time.
No. Transcription runs on your Mac with a local model, and there is no upload step in the product.
No. Speaker diarization is not in this beta, so turns are not labelled automatically. The transcript preserves the order of speech for manual review.
Up to 50 files per batch, 2 GB per file and 5 GB per batch. Start another batch when it finishes.
On the FLEURS English set (194 clips) the default model scored 5.93% word error rate. Unscripted conversation with crosstalk, accents or room noise is harder — the accented-speech set scored 19.18%. Check names and quotations against the audio before publishing.
mp3, wav and m4a, English only in this beta. Video files are not supported directly.
Beta users get free transcription for good. Leave your email to reserve the beta — no payment, no card. We'll send the signed build to the same inbox once it's ready. No newsletter, no drip sequence, no sharing your address.
We sent a confirmation link. Click it to confirm your address — your reservation is complete, and the beta build goes to the same inbox the moment it's signed.