What it does
Arcmira describes itself as a search engine for the spoken web. It indexes podcasts, interviews and long-form YouTube videos and organises what is said into people, organisations, products, topics and channels. You can search for a topic and get back spoken passages, each with a link to the exact moment in the video.
It also gives you a single video's transcript on request. The basic version is the video's own caption track, with a timestamp on every line. On paid plans, a Premium transcript is made from the audio and labels each line by speaker.
Beyond search, you can follow a person, company or topic and get alerts when they come up in newly indexed episodes. The same data is available on the website, through an HTTP API and through an MCP server for AI assistants such as Claude, ChatGPT, Cursor and Codex.
Who it's for
Writers, researchers, journalists and content teams who source material from podcasts, interviews and YouTube.
Key features
- Timestamped YouTube transcripts: the video's caption track on any plan, or a Premium transcript made from the audio on paid plans
- Speaker labels on Premium transcripts. Speakers show as Speaker 1, Speaker 2 and so on until a voice has been identified, and named speakers carry a confidence value
- Search for spoken passages across all indexed YouTube and podcast transcripts, filtered by channel, person or date, with a timestamped watch link for each result
- Mention research: every indexed moment a person, company, product or topic came up, with the video, start time and sentiment
- Public pages for people, organisations, products, topics and channels
- Monitors that alert you to new mentions by email or Slack, with webhooks on paid plans
- Arcmira Chat for asking questions in plain language
- An HTTP JSON API, a TypeScript SDK and a command-line tool
- An MCP server for Claude, ChatGPT, Cursor and Codex, with its source on GitHub
- Sponsor and ad-read data (Arcmira calls it commercial intelligence) on Pro+ and above
Based on Arcmira's website, pricing page and developer documentation, checked October 9, 2026. Features change, so check the official site.
Pros and cons
- Searches across a whole indexed library of conversations, not only the one video you paste in
- Every result links to a timestamp, which makes quotes easier to check and cite
- A free plan with 1,000 credits a month and no card, according to its pricing page
- Caption transcripts are cheap in credits: 4 credits per started 15 minutes of video
- The web app, API and MCP server share one account and one credit balance, and every plan includes the API and the MCP server
- Repeating the same request within 7 days costs nothing, according to its documentation
- It only covers what Arcmira has indexed. Its own documentation warns that an empty search does not prove a show is not covered, so it can't confirm that something was never said
- The Free plan withholds the newest 30 days of media and returns only the first page of a list, which limits it for current events
- Speaker-labelled Premium transcripts need a paid plan. They cost 300 credits per started 15 minutes of the whole video, even if you only need a short section
- Caption transcripts can be YouTube's automatic captions, which Arcmira notes often misspell names
- Speaker names can be inferred from context, such as an on-air introduction, so check who said a quote before you publish it
- Credit-based pricing takes some working out, and the most flexible features (API, MCP, webhooks) are aimed at technical users
- We found no option to upload your own audio files. To transcribe your own recordings, an editor such as Descript is a closer fit
Practical use cases
- Finding what a named expert or founder has said about a topic across many podcast appearances, with links to the exact moments
- Pulling a timestamped transcript of a long interview so you can quote it and link the source
- Checking whether a company, product or person has been discussed on podcasts, and roughly when and how often
- Setting a monitor on a beat, a client or a competitor so new mentions reach you by email or Slack
- Giving an AI assistant access to sourced spoken material through MCP for a research brief, then checking every quote against the original video
Pricing
Pricing checked October 9, 2026. Prices and plan details can change.
- Free: $0, 1,000 credits a month, 2 monitored names, no card. Withholds the newest 30 days of media and starts no new Premium transcripts
- Pro: $20/month, 20,000 credits a month (Arcmira says about 16 Premium transcript hours), 10 monitored names. 7-day free trial, card required
- Pro+: $60/month (first month $30), 60,000 credits a month, 30 monitored names, sponsor and ad-read data
- Ultra: $200/month (first month $100), 400,000 credits a month with up to 80 Premium transcript hours, 200 monitored names
- Enterprise: custom pricing
Yearly billing is 20% less: $192, $576 and $1,920 a year for Pro, Pro+ and Ultra. Credits reset monthly, and a team on any plan shares one balance.
What credits buy: a search result costs 4 credits (the first 5 in each search are free), a caption transcript 4 credits per started 15 minutes, a Premium transcript 300 credits per started 15 minutes (charged once per video per account), and a monitor alert 25 credits. On paid plans, extra usage costs $0.002 a credit up to a monthly limit you set. Check Arcmira's official pricing page for current details.
How it compares to alternatives
A basic transcript downloader gives you the captions for one video you already know about. Arcmira starts from the other end: you search for a topic, person or company and it finds the moments across the videos and podcasts it has indexed, links each one to a timestamp and can keep watching for new mentions. It also offers speaker-labelled transcripts, which caption files don't have.
It is not a recording or editing tool. If you need to transcribe and edit your own interviews or podcast episodes, Descript is closer to that job. For where research fits in a wider process, see our beginner's guide to an AI-powered writing workflow.
Editorial view: Arcmira is worth a look if your research regularly depends on what people say on podcasts and YouTube, and you want to find those moments by topic or person rather than watching hours of video. The free plan is enough to try the search and caption transcripts. Pro is the realistic starting point for current material and speaker-labelled transcripts. Treat it as a way to find and locate sources, not as a final authority: check quotes, speaker names and context against the original video before you publish.
This overview is based on Arcmira's published product information, pricing page and developer documentation, checked October 9, 2026. It does not describe hands-on testing.
Want to try it yourself?
The best way to judge fit is to test it on a real piece of your own work.