--- url: 'http://workflows-of-ai.com/blog/transcribe-audio-and-summarize' title: 'Transcribe audio and summarize' author: name: 'Marcus Johansson' url: 'http://workflows-of-ai.com/about/marcus-johansson' date: '2023-11-01T14:32:18+01:00' updated: '2026-10-10T12:14:04+02:00' type: article summary: 'Transcribe MP3 files with the AI Interpolator Deepgram module and summarize the transcript with OpenAI in Drupal, without writing code.' tags: - 'Marcus Johansson' - Workflows - 'Drupal AI' image: 'http://workflows-of-ai.com/sites/default/files/previews/57ff661575cee780.jpg' published: true og: site_name: 'Workflows of AI' 'image:width': '1280' 'image:height': '720' 'image:alt': 'Transcribe audio and summarize' --- Drupal AI # Transcribe audio and summarize Upload an MP3 and get a Deepgram transcription plus a two sentence OpenAI summary of what the speaker is thinking. Marcus Johansson November 1, 2023 ยท 2 min read ## About this video In this workflow, the process begins with an mp3 file featuring Damian Lewis's rendition of Antony's famous monologue from "Julius Caesar": "Friends, Romans, countrymen." This audio file is first uploaded into the system, where the AI Interpolator Deepgram module takes charge. Using its advanced audio processing capabilities, Deepgram transcribes the spoken words from the audio file, converting Antony's impassioned speech into a text format. The monologue can be found here: [https://www.youtube.com/watch?v=q89MLuLSJgk&ab\_channel=GuardianCulture](https://www.youtube.com/watch?v=q89MLuLSJgk&ab_channel=GuardianCulture) Once the transcription is complete, this raw, transcribed text is then fed into the AI Interpolator OpenAI module. The OpenAI algorithms, equipped with understanding the context and nuances of the text, proceed to distill Antony's lengthy and layered speech into a concise summary. This summary isn't just a shortened version, but a distilled representation of the core essence and meaning behind Antony's words, capturing what he truly intends to convey with his monologue. The outcome provides users with a quick and clear understanding of the monologue's primary message and sentiment. The result is: > The person is thinking about the funeral of Caesar and the conflicting opinions of his ambition. They express their loyalty to Caesar and question the judgment of those who criticize him. *This video is from before the AI Interpolator module became AI Automators, part of the [AI module](https://www.drupal.org/project/ai). Module and menu names have changed since, and some steps look different today.* ## Watch the video [Watch on YouTube: Transcribe audio and summarize](https://www.youtube.com/watch?v=ifXQDbMQMJY) ### Video chapters - 0:00 What Deepgram does - 0:57 Content type and audio field - 2:00 The Deepgram transcribe field - 2:35 A summary with OpenAI - 4:10 Testing on Julius Caesar - 5:05 Ideas for voice input ## Transcript YouTube's automatic captions, lightly corrected. Click a line to jump to that moment in the video. ### Transcript - 0:01 hey everyone uh Marcus here again with another workflow uh in this workflow I'm going to introduce the Deepgram module so Deepgram is a a service that can basically take a audio file uh and transcribe it - 0:16 for you in text so what we're going to do today is we're going to take we're going to set up so we can summarize uh uh a MP3 file basically what's being said in MP3 file - 0:31 uh as usual what you see in front of you is uh Drupal with the AI Interpolator modules installed um everything is open source everything is free you're free to use it after you - 0:43 have looked at this um in the future there's going to be in the video description some links to videos that uh we show you how to set up Drupal as well so off we go - 0:57 we start by logging in and we're just going to create that entity type for this as always it doesn't have to be nodes uh usually you work with custom entities but in this case we just do it - 1:12 so I can showcase everything over the GUI uh uh what do we call it summary of audio file and we just save this we don't need the body field that comes with Drupal by - 1:32 default um so the first thing we're going to do is we're going to create a file field that is the audio right so we're going to upload the audio file into there it's just going to be one you - 1:45 could do multiple and it would create multiple text fields of course only allowed MP3 you can do wob files as well you can do AAC files as well but in this case we take - 1:57 MP3s um and then then we're going to just add a text plane long that we call transcription so this is where the transcription will end up after we have uh uploaded right and for this to happen - 2:14 I have to enable the AI Interpolator and choose the rule uh Deepgram transcribe and we base this on the audio right I'm going to set up so it's running in batch mode and Save so now - 2:30 it's going to take that audio and transcribe it to this field but we want to take it one more step we want to summarize it right so we create another text long that we call - 2:43 summary um and that's just just be one field as well once again enable the AI Interpolator choose the OpenAI string long so we're going to take a text field and do a prompt on how to create uh - 2:58 another text field right and we're basing this on the transcription So based on the transcription create a two sentence summary of what the person talking about uh what the person - 3:24 talking is thinking so what I'm going to use uh is uh MP3 uh that comes from uh Damian Le is talking uh about um asking leer basically so anyway we need to set the - 3:51 context so the transcription and the context then comes from transcription here right from that field and here it's important that we first of all do batch mode but we also set the higher - 4:03 interpolation weight so the transcription has to exist before we do the summary right and that's it then we just add a common uh content that's called summary of audio file I have to give a title - 4:20 here uh and then we're just going to take this here uh oh it's from Julius Caesar sorry so that's what we're going to do and this is then going to fill in first transcription with what uh Damian Lewis - 4:34 is saying in this piece and then a summary basically of what he said so we click save and first it uploads creates a transcribe and then it summary so we have here transcriptions from what he's - 4:54 saying right and then a summary the person thinking about the funeral say Cesar and the conflicting options opinions of his ambition um and that's basically it um - 5:10 it's very very simple but opens up a lot of possibilities where you can uh not just upload MP3 files because of course if you're on web you can uh you can of course allow the person to speak into - 5:25 the microphone for instance so if you want to make one of these um startups that I have seen that you know summarizes your ideas from talking you could do it with this so that's it thank - 5:42 you Filed under [Workflows](/blog?category=6) [Drupal AI](/blog?category=2) Parts of this content were produced by AI. This site has not assessed whether a label is required for this content. How was AI used?The video and the introduction are Marcus's own, first published on workflows-of-ai.com. AI wrote the subtitle, the summary and the video chapters from the transcript, and corrected misheard names in YouTube's automatic captions. Marcus Johansson Tinkers with AI and Drupal [About Marcus](/about/marcus-johansson)[Drupal.org profile](https://www.drupal.org/u/marcus_johansson)[Drupal AI](https://www.drupal.org/project/ai) ## The post in three sentences Deepgram is a service that turns audio into text. A content type gets an MP3 field, a transcription field with the Deepgram transcribe rule, and a summary field where OpenAI writes two sentences from the transcription, with a higher weight so it runs second. It is tested on Damian Lewis reading Antony's speech from Julius Caesar.