Put the line in quotation marks and search the web for it. That finds any quote that made it into an article, a wiki, a subtitle site, or somebody's post, which covers most memorable lines from popular shows. If nothing comes back, one of two things is true: you are misremembering the wording, or the show is not well indexed in text. The next step for both is a subtitle search engine, which indexes caption tracks line by line instead of indexing pages about the show. Below is the order to try things in and how to tell when to stop.
Search the rarest fragment, not the whole line
Whole-sentence searches fail on one wrong word. Quotes drift in memory: articles get dropped, contractions get expanded, two lines get merged into one. Pick the three or four most unusual consecutive words in what you remember, put those in quotes, and leave the rest out. An unusual noun paired with an unusual verb beats a long accurate-sounding sentence almost every time.
Adding one piece of context helps: a character name, the word "episode", the decade, or the network. Adding three pieces usually hurts, because each one is a chance to be wrong about something the index actually has right.
Subtitle indexes are a different kind of search
A normal web search finds pages that talk about a show. A subtitle search finds the moment inside the show, because the underlying index is the caption track, timecode by timecode. Several sites do this over caption files and over public video platforms, and they behave differently from a search engine: they reward exact phrasing and punish paraphrase, and they usually return a timestamp you can jump to.
That timestamp is the part worth having. Finding out the show is one thing, and being able to watch the eleven seconds around the line is what most people were actually after. The same technique works on your own material, and searching by what was said is the general version of this problem.
When you are misquoting it
A large share of famous quotes are collectively misremembered. If exact search keeps failing, search the paraphrase without quotation marks and add the word "quote" or "misquote". Fan wikis and quote databases often list both the popular version and the actual line, which resolves it in one step and also explains why the exact search was never going to work.
When a person beats an index
Some things text search cannot do: a line delivered in a language you do not read, a show that never got captioned, a quote you only remember as a feeling. Communities built around identifying media handle these well. Give them everything peripheral you can: roughly when you saw it, where, the aspect ratio, whether it looked shot on film or video, an accent, a piece of set dressing. Peripheral details are what make identification possible when the words fail.
Audio matching apps are worth skipping here. That technology is built for music, matching a recording against a fingerprint of the same recording. Spoken dialogue that you are remembering rather than playing has no fingerprint to match.
When the quote is from something only you have
The methods above all depend on someone else having indexed the material. If the line was said in your own podcast archive, your own footage, or a recorded call, no public index contains it, and the search you want is over your own transcripts. People solve this with site-restricted web searches over posted transcripts, with grep over a folder of text files, or by remembering roughly which month it was and scrubbing. Anyone maintaining a few hundred episodes with transcripts knows how badly all three scale.
Vivu handles that side: it indexes what is inside the material you upload to a project, so "which episode did we talk about the pricing change" comes back as time ranges with the surrounding context, across every episode in that project at once instead of one file at a time. Episodes you upload later land in the same project and get searched with the rest.
If you are hunting through your own archive because you want to cut something out of it, the job is a bit different from identifying a quote, and working from the transcript to a usable clip has its own steps. If the target is a moment worth publishing rather than a specific line, selection works on different criteria entirely.
When to stop
Common phrases are unfindable. If the line is "we need to talk" or "I never wanted this", no index will help, because thousands of shows contain it and none of them are the one you mean. The same is true of lines you remember only in translation, or lines you may have actually heard in a film rather than a show.
Give it fifteen minutes across two methods. Beyond that, the expected return drops fast, and asking a community with good peripheral detail is a better use of the next hour than a sixth search phrasing.
Which problem you have
The split is simple. If the words were spoken in something the public has indexed, you have a phrasing problem, and the fix is a shorter, stranger fragment or a subtitle index. If the words were spoken in something only you hold, you have an indexing problem, and no amount of clever searching fixes it until the material itself is searchable.
FAQ
How do I find a quote when I only remember it roughly?
Search the two or three most unusual words separately, without quotation marks, and add one context word like a character name or genre. Exact-phrase search fails on a single wrong article or contraction, and rough memories are almost always wrong at that level of detail. If you get nothing, try searching the paraphrase together with the word "quote", since fan wikis frequently list both the misremembered version and the real line.
Can I search inside a show's dialogue rather than pages about the show?
Yes. Subtitle search engines index caption tracks line by line and return the episode and a timestamp, which is a different index from what a general web search uses. They work well for shows with widely available captions and poorly for anything that was never captioned or was only released in one region.
Is there a way to identify a show from a video clip instead of a quote?
Reverse image search on a distinctive frame is the fastest option, and it works better on a shot with an unusual set, costume, or on-screen text than on a close-up of a face. Failing that, communities that identify media from screenshots are effective when you give them the peripheral details: the aspect ratio, the picture quality, roughly when and where you saw it.
Why can I not find a quote from my own podcast or videos?
Because nothing outside your own files has indexed it. Web search covers pages that exist publicly, and a transcript sitting in a folder or pasted into a page that search engines crawl inconsistently is not the same as a searchable index. The practical fixes are keeping transcripts in one searchable place, or indexing the audio and video themselves so a phrase returns a timestamp instead of a filename.