The short answer
Upload your podcast episode (audio or video, up to 5 GB / ~6 hours) to VexaScribe and get a multi-speaker transcript with timestamps in ~10 minutes per hour of audio. Speaker labels work best for 2–4 voices. Per-hour cost ranges from $0.20 on Studio ($20/mo) to $0.60 on Starter ($2/mo); first 30 minutes free on signup.
Other tools worth knowing about: Descript if you also want a podcast EDITOR in the same tool (different product category — they own that). Riverside if you also need to record remote interviews ($24+/mo bundles both). Rev human transcription for ~99% accuracy if you can afford ~$90/episode for legal/journalism-grade work. Whisper local install if you have a GPU and want $0 unlimited.
Are You Transcribing Your Own Podcast or Researching Someone Else's?
These are two fundamentally different jobs — most transcription guides treat them as one. The output you want and the workflow that follows depend on which side you're on.
🎙️ My own podcast
You record episodes and need transcripts as raw material for downstream content.
- Show notes for your website (curated highlights + chapter timestamps)
- Blog post version of the episode (SEO + new audience)
- Quote extraction for Twitter/LinkedIn/email newsletter
- Searchable archive across episodes (find “harassment policy” across 100 episodes)
- Accessibility (~15% of US adults have some hearing loss per CDC)
🔍 Someone else's podcast
You're researching, analyzing, or sourcing material from episodes you didn't produce.
- Academic research (qualitative analysis of media content)
- Journalism (sourcing quotes from on-the-record podcast interviews)
- Competitive intelligence (tracking what executives say on their own pods)
- Brand mention tracking (where is your company being discussed?)
- Sentiment analysis at scale across an industry's podcasts
For personal research, journalism, and academic use, transcribing someone else's podcast is generally fair use. For commercial republishing of the transcript, get permission from the creator.
Show Notes vs Transcript vs Summary (Three Different Outputs)
These three terms get used interchangeably but mean different things. Knowing which one you need saves time and produces better results.
| Output | Typical length (1-hr episode) | Used for | Who creates it |
|---|---|---|---|
| 📄 Transcript | 8,000–15,000 words (literal text) | SEO publishing, accessibility, research, content repurposing | VexaScribe (AI transcribes audio → text) |
| 📝 Show notes | 300–800 words (curated) | Episode description, listener navigation, link sharing | You (writing from the transcript) or AI assistant |
| 📋 Summary | 100–400 words (5-10 bullet points) | Email teaser, social caption, executive briefing | AI summary feature (built on top of the transcript) |
VexaScribe produces the transcript as raw material. For AI-generated summaries on top, see our transcript-to-summary tool. Show notes are something you (or an AI assistant) write FROM the transcript — the transcript is the raw material; show notes are the polished deliverable.
Why Publish Transcripts? The SEO Case Most Podcasters Miss
⚡ The honest math
Podcast audio is invisible to Google search by default. The only thing search engines can index is your episode title and description (usually 100–300 words). A 1-hour interview contains 8,000–15,000 words of indexable content if you publish the transcript. That's 30–100× more search surface per episode.
Pacific Content and Edison Research have repeatedly documented measurable organic search growth from publishing podcast transcripts:
- 2–5× organic search traffic for shows that publish full transcripts vs audio-only over 6–12 months
- Long-tail keyword discovery — listeners find episodes through unrelated searches because their specific topic was discussed mid-episode
- Accessibility audience expansion — the CDC estimates ~15% of US adults have some hearing loss; deaf and hard-of-hearing readers are an underserved market
- International audience — transcripts can be machine-translated; audio can't (easily). Multi-language transcripts open non-English audiences
- AI training data exposure — ChatGPT, Claude, Perplexity cite transcribed content; audio is invisible to them
Source: Pacific Content's research on podcast SEO; Edison Research's annual “Infinite Dial” and “Podcast Consumer” reports; CDC hearing loss statistics. Treat the 2–5× range as directional — your actual lift depends on episode topic, niche competition, and on-page SEO basics (H2 structure, internal linking, schema markup).
Multi-Host Accuracy — The Honest Reality
Speaker diarization (auto-detecting who said what) is hard. Marketing copy usually says “automatic speaker detection” without telling you how it actually performs at scale. Realistic accuracy from Whisper-based diarization (which VexaScribe uses):
| Speaker count | Typical format | Realistic label accuracy |
|---|---|---|
| 2 speakers | Solo host + 1 guest (most common interview format) | 95%+ |
| 3–4 speakers | Co-hosts + 1–2 guests | 90–95% |
| 5–6 speakers | Panel discussions, roundtables | 80–90% |
| 7+ speakers | Chaotic panels, town halls | Manual review needed |
Hardest cases for any tool (including ours):
- Same-gender voices with similar vocal range and tone
- Overlapping speech (people talking over each other)
- Remote-recorded guests with very different audio quality from host
- Background music or sound effects bleeding into voice tracks
Best practice for podcasters: after the first transcription pass, rename “Speaker 1”, “Speaker 2” → actual host and guest names. Save the named pattern as a template for future episodes with the same hosts. See our guide to Whisper diarization for technical depth.
Handling Long Episodes (1, 2, 3+ Hours)
Long-form has become standard — Joe Rogan, Tim Ferriss, Lex Fridman, Acquired, Conan O'Brien all run 2–4+ hour episodes regularly. Most free transcription tools cap at ~25 MB (roughly 30 minutes of audio) and break on long-form. VexaScribe processes long episodes as a single file with no splitting.
| Episode length | MP3 size (128 kbps) | Processing time | Fits VexaScribe's 5 GB cap? |
|---|---|---|---|
| 1 hour (typical interview) | ~55 MB | ~5–10 min | ✓ Easily |
| 2 hours (deep-dive interview) | ~110 MB | ~15–20 min | ✓ Easily |
| 3 hours (Rogan-format) | ~165 MB | ~25–30 min | ✓ Easily |
| 4–6 hours (rare deep-dives) | ~220–330 MB | ~35–60 min | ✓ Yes |
For video podcasts (1080p MP4), file sizes are 5–10× larger — a 3-hour video podcast can hit 1–3 GB. Still under the 5 GB cap, but if your video podcast routinely runs longer than 6 hours, consider compressing to 720p with Handbrake first (audio quality is what matters for transcription, not visual resolution).
Repurposing Playbook — One Transcript → Five Derived Outputs
The leverage of a podcast transcript is downstream content. Here are five concrete derived outputs from one 1-hour episode transcript, with realistic effort estimates.
1. SEO blog post
Transcript → AI-generated outline → manual polish → publish on your podcast site. ~1 hour of editing work per episode. Captures search traffic the audio alone can't.
2. Email newsletter teaser
Extract 3–5 best quotes + 2-paragraph hook from the transcript. Send to your list with a link to the full episode. ~20 minutes per episode.
3. Twitter/X thread
10–15 quote tweets from the most insightful moments. Each tweet links back to the episode timestamp. Drives social discovery for free. ~30 minutes per episode.
4. YouTube Shorts / TikTok / Reels clips
Timestamped transcript makes clip identification fast — find the 30–60-second moments worth standalone shorts. Each short captioned with VexaScribe's SRT export. ~1 hour per episode for 3–5 clips.
5. LinkedIn post (B2B podcasts)
1–2 minute video clip + key quote + call-to-action. B2B podcasts especially benefit from LinkedIn distribution where the buyer audience lives. ~30 minutes per episode.
Total derived content from one transcript: roughly 3–4 hours of post-production work yielding 5+ pieces of content across as many channels. The transcript is the bottleneck unlock — you can't do any of this efficiently without one.
ポッドキャストコンテンツを再利用
1つの文字起こしから複数のコンテンツピースを作成。各エピソードの価値を最大化します。
番組ノート
詳細なエピソード要約を作成
ブログ記事
エピソードを記事に変換
ソーシャル引用
タイムスタンプ付きの共有可能な引用を抽出
YouTube字幕
動画版用にSRTファイルをエクスポート
SEOコンテンツ
エピソードをGoogle検索可能に
文字起こしから番組ノートへ
Before
After
対応プラットフォーム
ポッドキャスト文字起こし:DIY vs VexaScribe
手動文字起こし
- ✗1時間のエピソードに4〜6時間
- ✗自動話者ラベルなし
- ✗タイムスタンプを手動入力
- ✗外注すると高額
- ✗コンテンツ再利用が遅れる
おすすめ: 時間のある完璧主義者に最適
VexaScribeを使用
- ✓1時間のエピソードを5〜10分で
- ✓ホスト/ゲストラベルを自動生成
- ✓タイムスタンプを生成
- ✓音声1時間あたり$0.20から
- ✓当日に番組ノートを公開可能
おすすめ: 毎週配信するポッドキャスターに最適
ポッドキャスト文字起こしの仕組み
エピソードをアップロード
ポッドキャストの音声または動画ファイルをアップロードしてください。MP3、WAV、M4A、MP4などに対応しています。あらゆるポッドキャストホスティングプラットフォームからのエクスポートに対応しています。
AIが話者をラベル付け
AIがエピソードを文字起こしし、異なる話者を自動検出します。インタビューでホストとゲストを区別するのに最適です。
エクスポートして再利用
番組ノート用にテキスト、ブログ記事用にDOCX、YouTube字幕用にSRT/VTTとして文字起こしをダウンロードできます。1つの録音から複数のコンテンツピースを作成。
ポッドキャスターがVexaScribeを選ぶ理由
ポッドキャストワークフロー専用の機能
話者検出
ホストとゲストを自動的に区別します。番組ノートや引用の帰属を簡単に正しく設定できます。
番組ノート対応
番組ノート、エピソード要約、ブログコンテンツに簡単に変換できるフォーマットで文字起こしをエクスポートします。
引用対応タイムスタンプ
すべての文にタイムスタンプがあります。オーディオグラムやソーシャルクリップ用に正確なタイミングで引用を抽出できます。
YouTube字幕
動画ポッドキャスト用にSRT/VTTファイルをエクスポートできます。YouTubeに直接アップロードしたり、動画エディタに追加したりできます。
当日公開
録音した当日に文字起こしして番組ノートを公開できます。文字起こしバックログはもうありません。
国際的なオーディエンス
99言語で文字起こしできます。正確な多言語文字起こしでグローバルなリスナーにリーチできます。
ポッドキャスト文字起こしに関するよくある質問
RSSフィードから直接インポートできますか?
はい、ポッドキャストのRSSフィードURLを貼り付けて、エピソードを直接選択・インポートできます。手動でダウンロード・アップロードする必要なし。
ホストとゲストは別々に表示されますか?
はい、VexaScribeには自動話者識別機能があります。システムが異なる声を識別・ラベル付け。エディタで話者名を変更できます(例:「話者1」を「山田」に)。
BGMのあるエピソードはどうなりますか?
AIがBGMから音声を分離できます。軽いBGMは通常問題なし。音楽が大きすぎる部分は精度が下がる可能性があります。
YouTubeビデオポッドキャスト用の字幕を作成できますか?
はい。SRTまたはVTTフォーマットでエクスポートして、YouTube Studioに直接アップロード。タイムスタンプは自動同期。
過去のエピソードを文字起こしできますか?
もちろん。過去のエピソードを個別またはバッチでアップロード。ファイルサイズやエピソードの長さに制限なし。アーカイブ全体を検索可能に。
ファイルサイズに制限はありますか?
VexaScribeは任意のサイズのポッドキャストファイルに対応—数分の短いエピソードから数時間の長編番組まで。
注意: 文字起こしの精度は、音質、話者数、話し方の明瞭さによって異なります。BGMが結果に影響する場合があります。