Podcast Production Automation

# Podcast Production Automation

1Overview

This workflow automates the end-to-end podcast production pipeline: capture raw audio, remove background noise, generate timestamped transcripts, edit and prepare clips for distribution, and publish to multiple platforms simultaneously. Designed for podcast creators and production teams, this pack reduces manual editing from 5 hours to 1 hour per episode and saves 20–25 hours per month across the full production cycle.

---

2What's in This Pack

1. Descript

What it does: Descript is a cloud-based audio and video editing platform that transcribes your recordings automatically and lets you edit audio by editing text—similar to editing a document in Google Docs. It also handles basic audio cleanup and exports to various formats for distribution.

Role in this workflow: Descript serves as your central recording and initial editing hub, capturing raw podcast audio, generating rough transcripts, and preparing edited audio files for distribution.

Documentation: Descript Getting Started Guide

• Descript's transcription accuracy is good but not perfect; use Whisper as a backup for critical sections or legal content.
• Descript does not have a public API, so automation with Make will be limited to manual exports and uploads.

---

2. Whisper

What it does: Whisper is an open-source speech-to-text (voice-to-transcript) model developed by OpenAI. It converts audio files into accurate written transcripts and is free to use, either through OpenAI's API or by downloading and running the software locally on your computer.

Role in this workflow: Whisper provides a free, highly accurate transcription backup and alternative to Descript's built-in transcription, especially useful if Descript's accuracy needs verification or if you want to avoid vendor lock-in.

Documentation: Whisper GitHub Repository

• Whisper requires either an API key from OpenAI (if using the API) or technical setup to run locally; it is not a simple web app like Descript.
• For most users, Descript's transcription will be sufficient; use Whisper only if accuracy issues arise or if you need an open-source alternative.

---

3. Krisp

What it does: Krisp is an AI-powered noise removal and audio enhancement tool that strips background noise (traffic, wind, keyboard clicks, etc.) from recorded audio in real-time or in post-production. It works as a browser extension, desktop app, or API integration.

Role in this workflow: Krisp cleans up raw podcast audio before it enters Descript, ensuring that background noise doesn't interfere with transcription accuracy or final audio quality.

Documentation: Krisp Documentation

• Krisp works best on consistent, steady background noise; highly variable or loud environments may still require manual editing in Descript.
• Krisp integrates via API with Make, allowing you to automate noise removal in your production pipeline.

---

4. Murf AI

What it does: Murf AI is a text-to-speech (TTS) platform that converts written text into natural-sounding AI-generated voice narration. You can choose from dozens of voices in different accents, languages, and tones, and export audio files ready for use in videos, podcasts, or audiobooks.

Role in this workflow: Murf AI generates professional narration for podcast intros, outros, sponsor reads, or short-form clips extracted from full episodes—eliminating the need to re-record these segments manually.

Documentation: Murf AI Getting Started

• AI-generated voices sound professional but may not match your personal podcast voice; use Murf for standardized segments (intros, ads) rather than main content.
• Murf integrates with Zapier and Make via webhooks, allowing you to trigger voice generation automatically when new episode transcripts are ready.

---

5. Make (Integromat)

What it does: Make is a no-code automation platform (also called workflow automation or integration-as-a-service) that connects different apps and services so they can communicate and share data automatically. Instead of manually copying and pasting information between tools, Make creates workflows (called "scenarios") that run on a schedule or when triggered by an event.

Role in this workflow: Make orchestrates the entire distribution pipeline—taking your finalized audio and transcript from Descript, triggering Murf AI for clip narration, and publishing to podcasting platforms, social media, and email simultaneously.

Documentation: Make Documentation

• Make cannot directly integrate with Descript (which has no API), so your Make workflows will begin after you manually export files from Descript.
• Make can send audio files to podcasting platforms (Spotify, Apple Podcasts, etc.) via their distribution partners like Anchor or Transistor.

---

3Prerequisites

• Descript account: Paid Starter plan ($12/month) or higher. Sign up at descript.com.
• Krisp account: Paid Pro plan ($10/month) recommended for unlimited processing. Free tier available at krisp.ai for testing.
• Whisper: No account needed if running locally. If using via OpenAI API, you will need an OpenAI account and API key (available at platform.openai.com).
• Murf AI account: Paid Pro plan ($10/month) or free tier for testing. Sign up at murf.ai.
• Make account: Free plan available for testing; Pro plan ($20/month) recommended for production workflows. Sign up at make.com.
• Podcast distribution account: You will need access to a podcast host or distributor (e.g., Anchor, Transistor, Podbean, or Buzzsprout) to publish episodes. Most offer free plans.
• Recording software: Descript can record directly, but you may also use Audacity (free), GarageBand (Mac), or any audio recorder that exports MP3 or WAV files.
• Browser or desktop app access: All tools support web browsers (Chrome, Firefox, Safari, Edge); Descript and Krisp also offer desktop apps for Mac and Windows.

---

4Setup & Integration Guide

6. Setting Up Descript

1. Go to descript.com and click Sign Up.
2. Create an account using email or Google/Apple sign-in.
3. Choose the Starter plan ($12/month) during onboarding.
4. Verify your email and log in to the Descript dashboard.
5. Click New Project and select New Podcast (or New Recording if recording in-app).
6. Grant browser permissions for microphone and speaker access when prompted.
7. Test your microphone by recording a 10-second sample and confirm audio quality.
8. In Settings > Transcription, verify that automatic transcription is enabled (it is by default).

• Descript exports finished audio as MP3 or WAV files. Download these files and upload them manually to Make workflows or directly to your podcast hosting platform.
• Use the Share > Export menu in Descript to download transcripts as .txt or .srt (subtitles) for use in social media or video clips.

---

7. Setting Up Whisper

Option A: Using OpenAI API (Recommended for automation)

1. Go to platform.openai.com and sign in with your OpenAI account.
2. Navigate to API keys in the left sidebar.
3. Click Create new secret key and copy the key to a secure location (you will need it in Make).
4. Whisper API calls cost approximately $0.02 per minute of audio; set spending limits in Billing > Usage limits if desired.

Option B: Running Whisper Locally (For advanced users)

1. Install Python (version 3.8 or higher) on your computer from python.org.
2. Open a terminal or command prompt and run: pip install openai-whisper
3. Download your audio file to your computer.
4. In the terminal, run: whisper your-audio-file.mp3 --output_format txt (replace your-audio-file.mp3 with your actual filename).
5. Whisper will generate a .txt transcript in the same folder as your audio file.

• If using the API, you will paste your Whisper API key into Make workflows to trigger transcription automatically when audio is uploaded.
• If running locally, copy the generated .txt file and paste the text into Descript to verify or enhance the transcript.

---

8. Setting Up Krisp

1. Go to krisp.ai and click Sign Up.
2. Create an account using email or Google sign-in.
3. Choose the Pro plan ($10/month) at checkout (or start with the free tier to test).
4. Verify your email and log in to the Krisp dashboard.
5. You have two options for using Krisp:

Option A: Krisp Web App

• On the dashboard, click Upload Audio File (or drag and drop an MP3/WAV file).
• Click Remove Noise and wait for processing to complete (typically 2–5 minutes).
• Click Download to save the cleaned audio file.

Option B: Krisp Desktop App

• Download the Krisp app for Mac or Windows from the Downloads section.
• Install and open the app.
• Click Sign In and use your Krisp account credentials.
• Open your audio editor (Descript, Audacity, etc.) and select Krisp as your microphone input; Krisp will remove noise in real-time as you record.

1. Test noise removal by uploading a 30-second audio sample and comparing the before/after.

• Export cleaned audio from Krisp as MP3 and upload directly to Descript for transcription.
• To automate noise removal in Make, use the Krisp API by adding an HTTP request module (see Make Setup below for details).

---

9. Setting Up Murf AI

1. Go to murf.ai and click Start Creating.
2. Create an account using email or Google sign-in.
3. Choose the Pro plan ($10/month) (or test the free tier first).
4. Verify your email and log in to the Murf AI studio.
5. Click Create a new project and select Podcast or Voiceover.
6. Type or paste your script (e.g., podcast intro, sponsor read, or episode summary) into the text editor.
7. Click the Voice dropdown and select a voice by language, gender, and tone (e.g., "English - Female - Professional").
8. Click the Play icon to preview your narration.
9. Adjust Speed (words per minute) and Tone (confidence, warmth, etc.) if needed.
10. Click Export and choose MP3 as the file format.
11. Click Download and save the audio file to your computer.

• To automate Murf AI in Make, create a Make webhook (see Make Setup below). When a new episode transcript is ready, Make will send the script to Murf AI, generate narration, and download the audio automatically.
• Save Murf AI exports as MP3 and upload them to Descript to mix with main episode audio, or directly to your podcast host for distribution.

---

10. Setting Up Make

1. Go to make.com and click Sign Up.
2. Create an account using email or Google sign-in.
3. Choose the Pro plan ($20/month) during onboarding (or test the free plan first; free includes 1,000 operations/month).
4. Verify your email and log in to the Make dashboard.
5. Click Create a new scenario.
6. Click the empty module in the center and search for a trigger (the event that starts your workflow). For this workflow, use:

• Google Drive trigger: "Watch Files" (monitors a folder for new audio exports from Descript)
• Webhook trigger: "Custom Webhook" (you can trigger workflows manually or from external apps)

1. Select your trigger and follow the prompts to authenticate (e.g., connect your Google Drive account).
2. Click the + icon to add a second module. Search for and add the tools you want to automate:

• HTTP module (for sending data to APIs like Krisp or Murf AI)
• Zapier module (to trigger actions in Murf AI, Slack, etc.)
• Google Sheets module (to log transcripts and metadata)
• Email module (to send episode notifications)
• Slack module (to post updates to team channels)

1. Configure each module:

• In an HTTP module, paste the API URL and authentication key (API key).
• In a Zapier module, authenticate with the third-party tool and map input fields (e.g., transcript text) to output fields.

1. Click Save and then Turn on to activate your workflow.
2. Test the scenario by uploading a test file or manually triggering the webhook; check the Execution history to confirm each step completed.

• Descript → Make: Export audio/transcripts from Descript to Google Drive or Dropbox, then trigger a Make workflow to watch for new files.

• Make → Murf AI: Add an HTTP module with this configuration:
• URL: https://murf.ai/api/v1/speech/synthesize (example; check Murf AI docs for current endpoint)
• Method: POST
• Headers: Add Authorization: Bearer YOUR_MURF_API_KEY (replace with your actual Murf AI API key from Settings > API Keys)
• Body: Set to JSON format and include: { "text": "[your script]", "voiceId": "[murf voice ID]", "format": "mp3" }

• Make → Podcast Host: Add a Zapier or HTTP module pointing to your podcast host's API (e.g., Anchor, Transistor). Upload the finalized MP3 and metadata (title, description, publish date).

• Make → Slack: Add a Slack module to notify your team when an episode is ready to publish. Configure the message to include episode title and a link to the audio file.

---

5Step-by-Step Workflow

1. Descript: Record your podcast episode directly in the Descript app, or upload an existing audio file (MP3 or WAV). Descript will automatically transcribe the audio within seconds to minutes depending on file length.

• Data handoff: Descript generates a synchronized transcript alongside the audio waveform. You can now see the full text on screen and edit audio by clicking and editing the text.
• Pro tip: Use the Show Speaker Names feature to label different speakers; this makes editing multi-person podcasts much faster.

1. Krisp: Before finalizing in Descript, optionally export the raw audio and run it through Krisp to remove background noise. Download the cleaned version and re-upload it to Descript to re-transcribe if needed.

• Data handoff: Export cleaned MP3 from Krisp, then upload to Descript to replace the original audio.
• Pro tip: Krisp works best on steady background noise (air conditioning, office hum); use it first, then use Descript to manually remove any remaining noise artifacts (mouth clicks, breath sounds).

1. Descript: Edit the transcript by removing filler words ("um," "uh," "like"), false starts, and tangents. As you edit the text, the audio edits automatically.

• Data handoff: The edited transcript and audio remain in Descript.
• Pro tip: Use Descript > Search & Replace to quickly remove common filler words across the entire episode.

1. Descript: Export the finalized audio as MP3 via File > Export > Audio Only. Download the file and save to a folder on your computer or to a cloud storage service (Google Drive, Dropbox, etc.) that is monitored by Make.

• Data handoff: MP3 file is now ready for distribution and transcription verification.
• Pro tip: Descript also exports the transcript as an SRT (SubRip) file—save this too for generating video clips with captions.

1. Whisper (optional verification step): If you want to verify or enhance Descript's transcription, upload the final MP3 to Whisper via the OpenAI API (or run it locally). Compare the two transcripts and update the Descript transcript if Whisper's version is more accurate.

• Data handoff: Whisper outputs a plain-text .txt file. Copy the text and paste it into Descript to update the transcript if needed.
• Pro tip: Whisper is most useful for technical terms, names, or jargon that Descript may misinterpret.

1. Murf AI: Create short clips or summaries of your episode (intro, key takeaway, sponsor read). Paste the text into Murf AI, select a voice, and generate MP3 narration.

• Data handoff: Murf AI exports MP3 files. Download these and store them alongside your main episode audio.
• Pro tip: Use the same voice and tone across all episodes for brand consistency; save your voice preferences as a Murf AI template.

1. Make: Set up an automated workflow to handle distribution. Create a scenario that:

• Watches for new MP3 files in your Google Drive/Dropbox folder (trigger: Google Drive > Watch Files)
• Downloads the audio and transcript
• Sends the episode title, description, and audio file to your podcast host (via API or Zapier) to publish across Spotify, Apple Podcasts, and other platforms
• Uploads Murf AI-generated clips to social media platforms (Instagram, TikTok, Twitter/X) using Zapier
• Posts a notification to your team Slack channel when the episode goes live
• Logs episode metadata (title, publish date, platform URLs) in a Google Sheet for tracking

• Data handoff: Make coordinates all downstream systems. The workflow receives MP3 + metadata from Descript and distributes to podcast hosts, social media, and team communication tools.
• Pro tip: Use Make's scheduling feature to publish episodes at optimal times (e.g., 6 AM on Tuesdays for maximum reach).

1. Podcast Host & Social Media: Your Make workflow publishes the episode automatically across Spotify, Apple Podcasts, YouTube, TikTok, Instagram, etc. No manual uploads needed.

• Data handoff: Each platform receives the MP3 file, show notes (transcript), and metadata (artwork, title, description) from Make.
• Pro tip: Use your podcast host's analytics dashboard to track listener growth and identify popular episodes; use this data to guide future episode topics.

---

6Integration Map

Here is how data flows through this workflow:

``` Raw Audio (MP3/WAV) ↓ [KRISP] → Cleaned Audio (MP3) ↓ [DESCRIPT] → Transcript (TXT/SRT) + Edited Audio (MP3) ↓ [WHISPER] (optional) → Verified Transcript (TXT) ↓ [MURF AI] → Narrated Clips (MP3) ↓ [MAKE] → Distributes to: ├→ Podcast Host (Anchor, Transistor, etc.) ├→ Social Media (Instagram, TikTok, Twitter/X via Zapier) ├→ Email (Newsletter) ├→ Slack (Team notification) └→ Google Sheets (Episode log) ```

File formats at each handoff:

• Raw recording → MP3 or WAV (Descript accepts both)
• Descript output → MP3 (audio) + TXT/SRT (transcript)
• Krisp output → MP3 (noise-reduced audio)
• Whisper output → TXT or JSON (transcript)
• Murf AI output → MP3 (narrated audio)
• Make distribution → MP3 + JSON metadata (to APIs) or MP3 + TXT (to social media as uploads)

Fully automated handoffs (via Make):

• Descript → Make (via Google Drive/Dropbox file monitoring)
• Make → Podcast Host (via REST API—HTTP request with authentication)
• Make → Social Media (via Zapier connector)
• Make → Email/Slack (via native Make modules)

Manual copy-paste steps (unavoidable due to tool limitations):

• Descript → Krisp (Descript has no API; export MP3 and upload to Krisp manually)
• Descript → Make (same reason; download transcripts and metadata, then paste into Make or upload files to Google Drive for Make to monitor)
• Whisper → Descript (if using Whisper for verification; copy/paste corrected text back into Descript)

---

7Troubleshooting

1. Open the Descript transcript and locate the error (the text will be out of sync with the audio).
2. Click on the incorrect word in the transcript and manually type the correct word.
3. (Optional) Export the audio to Whisper to verify the correct transcription; if Whisper is more accurate, copy its output and update Descript.
4. Re-export the corrected transcript as SRT or TXT for downstream use.

---

1. In the Krisp web app, click the dropdown next to Remove Noise and select Medium or Light instead of Heavy.
2. If using the Krisp desktop app, open the app settings and lower the Noise Suppression Level slider.
3. Re-process the audio and listen to a 30-second sample to confirm it sounds natural.
4. If noise removal still causes issues, skip Krisp and perform manual cleanup in Descript instead (use the Remove Sound tool on specific sections).

---

1. Verify that your API key for the podcast host is correct. Go to the host's Settings > API Keys and copy the key again.
2. In the Make HTTP module, paste the API key in the Headers section in the correct format (e.g., Authorization: Bearer YOUR_API_KEY).
3. Check the podcast host's API documentation to confirm the correct endpoint URL (e.g., https://api.example.com/episodes/create).
4. Test the endpoint using a tool like Postman (postman.com) before running the full Make workflow.
5. In Make, click Run once to test the scenario and check the Execution history for specific error messages.

---

1. In the Murf AI editor, reduce the Speed slider from the default (usually 1.0) to 0.8 or 0.9 words per minute to sound more natural.
2. Increase the Pause duration between sentences (usually 0.5 seconds) to allow the voice to "breathe."
3. Try a different voice: click the Voice dropdown and audition 2–3 alternative voices with different tones (e.g., "Warm" vs. "Professional").
4. Re-generate and download the narration as MP3.

---

1. Verify that your social media account is properly authenticated in Make. Go to Make > Your Connections and check that Instagram, TikTok, or Twitter/X shows a "Connected" status.
2. If disconnected, click the connection and re-authenticate by signing into your social media account.
3. Check that your Make scenario includes the correct social media module (e.g., Instagram > Create a Post) with the right input fields (image/video file, caption, hashtags).
4. If scheduling posts, verify that the Schedule date and time are set correctly in the module and that your computer's time zone is correct (Make uses your account's time zone).
5. Test by posting a single image manually through Make; if it succeeds, the full workflow should work.

---

1. Confirm that your Google Drive > Watch Files trigger is monitoring the correct folder. In the trigger settings, click Select a Folder and choose the Google Drive folder where you save Descript exports.
2. Verify that your Google account is properly authenticated in Make. If authentication failed, click Reconnect in the module and re-sign into Google.
3. Manually test the trigger: upload a test MP3 file to the watched folder, then in Make, click Run once to force the scenario to check for new files.
4. Check the Execution history in Make to see if the file was detected; if not, verify that the folder path is correct and that you have read/write permissions to that folder.
5. If issues persist, use Zapier as an alternative automation platform (supported by most tools in this pack) to trigger workflows instead of Make.