Urgent.News

What's breaking now, across thousands of outlets.

Tech

Como transcrever e resumir vídeos do YouTube automaticamente com Google Apps Script e Supadata

No iadots.com, eu reúno vídeos sobre IA, agentes, robótica e humanoides em quatro idiomas. Assistir a cada vídeo e escrever um resumo à mão não escala, então montei uma automação simples: um script no Google Apps Script pega a transcrição do vídeo pela API da Supadata, manda o texto para um modelo de linguagem e salva o resumo. Neste tutorial mostro como montar a mesma estrutura do zero, com o…

Original Portuguese Read in English

This article provides a step-by-step guide on how to automatically transcribe and summarize YouTube videos using Google Apps Script and Supadata. The author, Dev.to, explains that manually transcribing and summarizing videos is not scalable, and they created an automation using a Google Apps Script to retrieve the video transcript via Supadata's API, send the text to a language model, and save the summary.

The tutorial walks through the process of setting up the project, including obtaining API keys for both Supadata and Gemini, creating a Google Sheet with a "Videos" tab and three columns (youtubeId, lang, and resumo), and writing the necessary code to fetch the transcript, convert it to text, and generate a summary using a language model.

The article emphasizes the importance of securely storing API keys and handling potential issues such as videos with long durations or incorrect language requests.

Brief written by urgent.news from Dev.to's own syndicated text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in Tech

A good prediction doesn't always make a good decision

I started building an FPL tool because I kept coming back to one question: what's the point of predicting the best player if signing him isn't actually the best move?

  • Predicting best player doesn't guarantee best decision
  • FPL Decision Engine merges data, AI, rules to evaluate real decisions
  • Engine's effectiveness in improving FPL performance still under testing

From One Prompt to 39 Finished Products: What Actually Broke

Last March I ran a batch of 40 uploads to Adobe Stock overnight. I woke up to a rejection screen and a bad feeling. The art was fine. The CSV was not.

  • Single prompt generated 39 finished products
  • Metadata mismatch caused initial 40 uploads rejection
  • Scripts automate metadata and upload tracking

The License Layer — Or, Who Can Rewrite the Terms Underneath You?

Twelve essays into this series, the line has never moved: the layer you don't own is the layer that owns you. I've named distribution (#45), the model (#46), identity (#47), access (#48), the harness…

  • License Layer controls software use, modification, distribution
  • Bitwarden switched to dual-license model, sparking debate
  • Nvidia acquisition, Apple removal highlight license control risks

More from Monday 12 October →