Podcast Transcription

Upload your podcast episode file (MP3, WAV, M4A, or other supported formats) to VexaScribe (formerly NovaScribe). The AI automatically transcribes the audio using OpenAI's Whisper Large-v3, detects different speakers (host, co-host, guests), and generates word-level timestamps. A typical 60-minute episode completes in 5-10 minutes.

Yes. VexaScribe includes automatic speaker diarization. When your podcast has multiple speakers, the system labels each one separately (Speaker 1, Speaker 2, Speaker 3...). You can rename them in the editor ("Host", "Guest: Jane Smith", "Co-host: Dan") and the labels carry through every export. Accuracy is highest for 2-3 speakers on separate microphones; degrades with overlapping speech and 4+ speakers.

Turn one episode into multiple pieces of content: create show notes for your website, repurpose as blog posts (1,500-2,500 words from a 45-60 min episode), extract quotes for social media, generate YouTube captions (SRT/VTT export), improve SEO with searchable text, and make your content accessible to deaf and hard-of-hearing listeners.

A typical 1-hour podcast episode is transcribed in about 5-10 minutes. Processing time depends on audio length, quality, and current load. You can close your browser while processing — the transcript will be waiting in your dashboard when it's ready.

VexaScribe supports all common podcast formats: MP3, WAV, M4A, FLAC, OGG, AAC, AIFF, WMA, AMR, OPUS for audio. Video formats (MP4, MOV, AVI, MKV, WebM) work too — we extract the audio automatically. Files can be up to 5 GB and 10 hours per upload, which covers virtually any podcast episode.

Yes. VexaScribe supports 99 languages for transcription with automatic language detection. You can also translate transcripts to any of 133 languages using the built-in translation feature — useful for international podcasts that want English subtitles, or English podcasts reaching non-English audiences.

VexaScribe plans start at $2/month for 200 minutes. A weekly 1-hour podcast (~4 hours/month) costs less than $3/month on the Starter plan. Compared to human transcription services like Rev ($1.50/min, ~$90/hr) or Happy Scribe ($1.70/min), AI is 150-450× cheaper. For most podcasters, AI accuracy at ~95% on clear audio is more than enough.

Yes. VexaScribe exports SRT and VTT subtitle files with word-level timestamps. Upload directly to YouTube using "Add language → Upload file with timing" → select your .srt file. Same files work in Vimeo, Spotify Video, or any video editor (Premiere Pro, Final Cut Pro, DaVinci Resolve).

Apple Podcasts launched auto-generated transcripts in March 2024 with iOS 17.4 — but only for English, French, Spanish, and German. Auto-transcription rolled out to new episodes first, with the back catalog filling in over time. If you publish in another language, or want to control the exact transcript shown, you'll still need to provide your own. Spotify also added optional creator-uploaded transcripts (VTT/SRT) starting in 2023-24, but it's opt-in, not automatic for all podcasts.

On clear podcast audio with 2-3 speakers on separate microphones, AI transcription accuracy is around 95% (5% Word Error Rate). Speaker diarization is accurate when speakers have distinct voices and don't talk over each other. State-of-the-art research shows pyannote 3.1 achieves 11-19% Diarization Error Rate on real-world audio. Conditions that hurt accuracy: overlapping speech, similar voices, 4+ speakers, low-bitrate audio.

Yes. Chapter markers with timestamps are one of the outputs when you generate an AI summary from your transcript — pick the Podcast summary type in the summary editor to get a chapter list with times, per-chapter descriptions, and key quotes. Apple Podcasts and Spotify both accept chapter markers via ID3 chapter tags embedded in the episode audio file. Most podcast hosting platforms (Buzzsprout, Transistor, Captivate, Simplecast) let you paste chapter timestamps directly into the episode edit page rather than encoding them into the audio. Chapters help discovery on both platforms and reduce listener drop-off.

Yes. VexaScribe supports bulk transcription — upload up to 50 episodes at once as a single batch, and download all transcripts plus AI summaries as a ZIP archive when the batch completes. This is the workflow most podcast agencies and prolific podcasters use to catch up on 6-12 months of un-transcribed back catalog in one session. Batch processing uses the same $2-$20/month plan minutes as single uploads. For very large backlogs (100+ episodes), split into multiple batches; the Studio plan at $20/month covers about 100 hours of audio.

Yes, in two steps. First, get the transcript. Second, use the Podcast summary type to generate structured show notes with chapters, key quotes, and speaker discussion points — that's already usable as a blog post skeleton. For a longer piece, use the transcript editor's regenerate feature with the Lecture or General summary type to get a different structural angle, then paste both into your editor of choice. A typical 45-60 minute episode produces 1,500-2,500 words of blog content. Publishing the transcript itself alongside the episode also has meaningful SEO impact — searchable long-tail keywords the audio can't rank for on its own.

Five methods, honest order. (1) Apple Podcasts — since March 2024, most English/French/Spanish/German shows have auto-transcripts. Open the episode, tap the quote-bubble icon at the top-right. Free, no signup, viewer-only (no download). (2) Spotify — creators can opt-in to display transcripts on the episode page; coverage is thin outside major shows. (3) HappyScribe podcast library (podcasts.happyscribe.com) — user-uploaded searchable transcripts, mostly popular English shows. (4) YouTube — if the podcast publishes to YouTube, use /tools/youtube-transcript to pull the auto-caption. (5) Upload the audio yourself to VexaScribe — works for any podcast, any language, produces a downloadable transcript in TXT/DOCX/SRT with speaker labels. Method 5 is the only path that reliably works for non-English shows, obscure podcasts, and any use case that needs the transcript as a file.

Four places. (1) Apple Podcasts app — tap the quote-bubble icon on episodes that have auto-transcripts (mostly English/French/Spanish/German post-2024). (2) Spotify — some shows opt-in to display transcripts on the web episode page; check the episode's page on open.spotify.com. (3) HappyScribe's podcast library (podcasts.happyscribe.com) — searchable database of user-contributed podcast transcripts, tilted toward popular English shows. (4) The podcast's own website or Substack — some hosts publish transcripts for SEO or accessibility. If none of these have your show, upload the episode audio here and get a full transcript back in 5-10 minutes.

Partially, since 2023. Spotify supports creator-uploaded transcripts (VTT/SRT) that display on the web episode page and inside the app on some devices. It's opt-in, not automatic — the creator has to upload the file — so coverage is thin outside major podcasts. Spotify's own official podcasts have transcripts; smaller independent shows usually don't. If Spotify doesn't show a transcript for the episode you want, download the audio (Spotify allows for offline listening on paid plans; RSS-published shows are downloadable directly) and upload here for a full AI transcript in your language.

Yes, for many shows since March 2024 (iOS 17.4). Apple auto-generates transcripts for episodes in English, French, Spanish, and German — no creator action required. Look for the quote-bubble icon (💬) in the top-right of the episode player; tap it to view the transcript. Limits: (1) viewer-only, no download or export; (2) English/French/Spanish/German only — other languages don't have auto-transcripts yet; (3) accuracy is Apple's proprietary model, comparable to Otter, not always as accurate as Whisper Large-v3. For non-Apple-supported languages, exportable files, or higher accuracy, upload the audio here.