Compact daily workflow (10-minute episode)

Goal: fast, repeatable pipeline that goes text → natural voice → light cleanup and batch export, done in under an hour once you’re set up.

Recommended workflow (works for one-person creators)
1) Script prep: write a 1,200–1,500 word script (≈10 minutes at 120–150 wpm). Use ChatGPT to generate a conversational draft and then tidy it: break into 60–90s chunks, add short sentences where you want natural pauses, and mark emphasis or explicit pause cues like “[pause 600ms]”.
2) TTS batch render (ElevenLabs): create a voice preset you like (speed, pitch/style). Export each chunk as a separate file so edits are local and cheap to re-render. If you automate, call the ElevenLabs API to batch-render segments and name them episode_01_01.wav, episode_01_02.wav…
3) Quick cleanup: import all segments to a simple editor (Descript, Audacity, or Reaper). Join in order, then run breath removal and pacing tweaks.
4) Finalize: apply mild compression, normalize to your target LUFS (-16 LUFS recommended for streaming/podcast consistency), export MP3/MP4 or WAV in batch.

Practical tips for breaths and pacing
- Reduce breaths at source: when writing, shorten interjections and use commas/periods to give natural space. In ElevenLabs, lower “breathiness” or choose a voice/style that has fewer audible breaths (test voices).
- Automated breath removal (paid): iZotope RX Breath Control or Descript’s filler/breath removal gives fastest clean results.
- Free/manual: in Audacity, zoom to breaths and silence or reduce gain; use a short noise gate (careful with strong settings) and then a light click/pop removal. Manual edits for a 10-minute show take 10–20 minutes.
- Pacing: if a sentence feels rushed or slow, split the audio and insert a 200–800ms pause or re-render that segment with a slightly slower/faster voice rate. Use time-stretch (change tempo) rather than pitch shift.

Batch export and naming
- Export per-episode file and a stereo master. Keep segment files until the episode is posted.
- File format: WAV 44.1 kHz / 16-bit for masters; MP3 128–192 kbps for delivery.
- If using ElevenLabs API, export as WAV and run your final processing locally to keep a simple version-control friendly folder (raw_tts/, edited/, masters/).

Recommendation
If you want the fastest recurring workflow and are comfortable with some tools: use ChatGPT for script drafts, ElevenLabs for voice generation, and Descript (or Audacity + iZotope RX if you prefer) for breath removal and final touches. That combo minimizes manual work while keeping quality high.

Decision criteria (pick based on these)
- Budget: free=Audacity + manual edits; low-cost=subscription Descript; premium=iZotope RX + ElevenLabs pro features.
- Skill level: beginner=Descript (transcript-driven edits); intermediate=Audacity/Reaper; advanced=API automation + RX.
- Output quality needed: standard podcast = ElevenLabs TTS + light cleanup; broadcast-level = custom voice profiling + RX breaths.

Quick checklist before export
- [ ] Script chunked into ~60–90s segments
- [ ] TTS presets saved and all segments generated
- [ ] Breath removal applied (auto or manual)
- [ ] Pacing adjusted; re-render segments where needed
- [ ] Compression and loudness normalization (-16 LUFS)
- [ ] Master exported and filenames/versioned

Best-for / Avoid-if
- Best for: solo creators producing daily short shows who need consistent, fast production.
- Avoid if: you need fully human nuance for interviews or extremely high-end broadcast quality (then consider human VO or heavy postproduction).

If you want, I can give a short script template tuned for ElevenLabs voice generation and a sample API batch command for automating exports.

Compare options for ElevenLabs