TranscriptFetch

About TranscriptFetch

TranscriptFetch is a transcript API built on the observation that most primary-source material now exists only as speech — the founder explaining the product on a podcast, the engineer demonstrating it on YouTube, the customer reviewing it on TikTok — and that text-first AI pipelines are blind to all of it. One API key covers five platforms: YouTube, TikTok, Instagram, Spotify and general podcasts. Post a video or podcast URL and it returns per-segment text with start times, using platform captions where they exist and AI transcription where they do not, so a downstream model can cite a specific timestamp rather than paraphrasing the whole thing. Alongside transcripts it exposes listing calls — search, channel and playlist — so an agent can walk a creator's whole back catalogue rather than being handed one URL at a time, and everything comes back in a single consistent response shape. The billing design is deliberately builder-friendly in two ways: only successful responses are billed, so a failed or empty fetch costs nothing, and every account gets a hundred free credits a month without a card. There is an MCP server and packaged agent skills from the Pro tier upward, so Claude or ChatGPT can pull and summarise a video directly in a conversation. The vendor publishes 99.82 percent API uptime and commits to a twelve-month deprecation notice, which is a more concrete reliability promise than most tools at this price make.

Key Features

Tags