An n8n workflow that transcribes interview audio locally with Whisper, then uses Claude to index questions and polish the transcript into a formatted Google Doc.
“N8N Interview Transcription Robot with Local Whisper and Claude Cleanup” is an intermediate Content AI agent workflow from Media. It connects n8n, Claude, Google Sheets and Other or custom in 5 steps. Setup takes about Not stated and running costs are about Not stated.
Turning raw interview recordings into a usable written record usually means manual transcription, then hand-scanning hours of text to find where each question was asked, then reformatting everything for sharing. This workflow automates the whole chain: audio goes in, a cleaned and question-indexed document comes out in Google Workspace.
Create an n8n workflow that starts when an interview .wav file is supplied. Keep the intake simple so a recording can be dropped in without extra processing, and pass the file reference along to the transcription stage.
Add a step that runs Whisper on the local machine rather than a hosted API. This keeps the audio on your own hardware and gives you a foundation you can extend into a fully local pipeline later if privacy requirements change.
Feed the raw transcript to Claude Sonnet and ask it to identify each interviewer question and where it appears. The output is a structured index that maps questions to their position in the conversation.
Run a second Claude pass, this time with Haiku, to tidy the transcript text: fix obvious transcription artifacts, normalize speaker turns, and prepare the body for reading. Keep the question index attached so the cleanup preserves structure.
Use the Google Workspace integration in n8n to create a document containing the cleaned transcript plus the question index. Format headings and sections so the result is ready to share without further editing.
Using n8n, build a workflow that accepts an uploaded [INTERVIEW WAV FILE], runs it through a local Whisper transcription step, then sends the transcript to Claude Sonnet to produce a question index listing each interviewer question and its location. After that, pass the transcript and index to Claude Haiku to clean up transcription artifacts and normalize speaker turns. Finally, create a formatted document in Google Workspace containing the cleaned transcript with the question index as a section. Return the workflow as importable n8n JSON and note any credentials or environment variables I need to configure.
You are processing an interview transcript. First, act as an indexing pass: identify every question asked by the interviewer and record the question text with its approximate position in the transcript. Then act as a cleanup pass: correct obvious speech-to-text errors, normalize speaker labels, and keep the original meaning intact without adding content. Output two sections, a question index and a cleaned transcript, formatted so they can be inserted into a Google Doc. Here is the transcript: [RAW WHISPER TRANSCRIPT].
An n8n workflow that transcribes interview audio locally with Whisper, then uses Claude to index questions and polish the transcript into a formatted Google Doc.
You need n8n, Claude, Google Sheets and Other or custom.
Plan for about Not stated. It is rated intermediate. The guide walks through 5 steps, starting with “Set up the n8n trigger and upload path”.
Expect about Not stated in running costs ($25–100/mo range).
Turning raw interview recordings into a usable written record usually means manual transcription, then hand-scanning hours of text to find where each question was asked, then reformatting everything for sharing. This workflow automates the whole chain: audio goes in, a cleaned and question-indexed document comes out in Google Workspace.
Score every demo request against your ICP, update HubSpot and ping the right rep in Slack within a minute of the form submit.
Every new Zendesk ticket gets a category, priority and sentiment tag plus a draft reply in an internal note before an agent opens it.
Export the Descript transcript, let ChatGPT pick the five strongest moments, and get 10 platform-ready posts in a review doc within minutes.
Built this workflow? Show it is featured on AgentUseHub with a live upvote count that links back here. About badges
