Copy a public link
In Instagram, open a Reel, choose Share, then Copy link.
FROM REEL TO READABLE
instagram transcript.
Instagram transcript generator · Reels to text, one link at a time.
01
Drop the link.
Text you can copy. A file you can keep.
Ready when you are.
02
Take the words.
Your words,
right here.
Nothing to copy yet
ONE LINK IN.JUST THE WORDS OUT.
THE RECORDED CHECK
We tested the first three Google results.
Two videos. Six checks.
Recorded on 3 November 2026.
We used each tool's free link-input path. Tool labels follow their order in that Google results session; these are recorded outcomes, not a ranking of quality.
| Tool | Video | Observed result | Access used | Input |
|---|---|---|---|---|
| Tool A | Video 1 | Text returned | No login, payment or installation | Paste link |
| Tool A | Video 2 | No text: result area empty | No login, payment or installation | Paste link |
| Tool B | Video 1 | Text returned | No login, payment or installation | Paste link |
| Tool B | Video 2 | No text: transcript not found | No login, payment or installation | Paste link |
| Tool C | Video 1 | No text: error returned | No login, payment or installation | Paste link |
| Tool C | Video 2 | No text: error returned | No login, payment or installation | Paste link |
Text was returned in two of the six checks.
Both tools that returned text did so for video 1.
None of the three free paths returned text for video 2; this does not establish why.
Those two outputs began with the same twenty words, word for word.
Video 1: DfQd6cYFMUd
Video 2: DcbpfDsjctR
These checks do not measure processing time, prove why a transcript was missing, or establish what a paid or signed-in path would return. A security verification appeared on one free path. This table records those tools, not this page's transcription results.
A SEPARATE RECHECK
We reran the same videos through a different API route.
Requests completed on 3 November 2026.
Credit balance checked on 4 November 2026.
We sent each of the same two public links once through a hosted API, using speech-generation mode on its free account. This is a separate check; the six recorded results above are unchanged. The tool at the top now uses a cloud API route; the dated observations in these two tables are unchanged.
| Video | Response and text | Words | First twenty words | Credits used |
|---|---|---|---|---|
| Video 1 | HTTP 200; text returned | 228 | Before we go, check this out. NASA getting ready for the first manned trip to the moon since 1972. A | 2 |
| Video 2 | HTTP 200; empty transcript | 0 | None returned | 2 |
Word counts split the returned text on whitespace. Video 1 returned text, but completeness and word-for-word accuracy were not independently verified. Video 2 returned no words and no restriction error. An HTTP success status alone does not mean a usable transcript was produced.
Video 2's complete response body:{"lang":"none","availableLangs":["none"],"content":[]}
Inference: the local all-zero recognition input and the API's empty transcript are consistent with no usable speech reaching recognition in these two attempts. They do not prove that the original Reel is silent or that both routes received the same audio file.
Account usage changed from 0 to 2 to 4 credits: Video 1 used 2 − 0 = 2, and Video 2 used 4 − 2 = 2. Both response headers also reported 2 billable credits. The free account had 100 − 4 = 96 credits remaining at the balance check. These two requests do not establish a fixed cost or an access guarantee for other videos.
ADDITIONAL RECORDED REQUESTS
What five more inputs returned.
Completed 3 November 2026 UTC.
Historical observations, not new requests.
For our downloaded 477.097551-second sample, independent listening and local recognition confirmed speech; the API returned 1024 whitespace-counted words and used 2 credits. DUTB7rKjeZ8
Our downloaded 50.062222-second silent sample contained all-zero decoded audio across the complete file; its API request returned 0 words but still used 2 credits. BRrTTH8hcTS
For our downloaded 53.124399-second sample, independent listening and local recognition confirmed speech; the API returned 136 whitespace-counted words and used 2 credits. C_iMQyhMY8s
For our downloaded 46.685375-second sample, independent listening and local recognition confirmed speech; the API returned 125 whitespace-counted words and used 2 credits. DL_Kt2APFXU
For our downloaded 160.075510-second sample, independent listening and local recognition confirmed speech; the API returned 343 whitespace-counted words and used 2 credits. Da0kJgBs3Bz
The response for our 477.097551-second sample contained 85 text segments, with the first starting at 800 milliseconds and the last starting at 468600 milliseconds with a duration of 5000 milliseconds.
In the 46.685375-second sample, the last returned text segment started at 35080 milliseconds and lasted 4480 milliseconds; that record does not verify the remaining audio.
For the 53.124399-second sample, the last returned segment started at 45320 milliseconds and lasted 1000 milliseconds; we did not certify a complete verbatim transcript.
Across these five requests, recorded account usage progressed from 4 to 6, 8, 10, 12 and 14 credits; each corresponding billing response header reported 2.
The archived series therefore consumed 14 minus 4, or 10 credits, leaving 100 minus 14, or 86 credits at that historical balance check.
We independently checked the four speech-positive samples against their downloaded audio and local recognition of the first 45 seconds, rather than treating returned text as proof of speech.
The silent sample was checked across its complete decoded audio, alongside its publisher's no-audio description; a transcript response was not used to establish silence.
All five archived API responses had HTTP 200, including the zero-word silent sample; the status alone did not distinguish a usable transcript from an empty one.
Those requests were made on the free account with generation mode and timestamped segment output, once per source, without specifying a language.
Two archived clips exceed this website's current two-minute limit; their direct API results do not mean this tool accepts clips of those lengths.
THE SHORT VERSION
Less watching.
More working with words.
Get the spoken text
Paste it above and press Enter. Keep this tab open while the audio is processed.
Take it with you
Copy the transcript or download a plain-text file. Check names and quotations against the audio.
Supported links, and the limits
Public Reel, single-video post and IGTV link formats are accepted. A transcript still requires an accessible audio track. The current processing limit is two minutes; longer clips are rejected rather than silently trimmed. File size is not measured by this route, and the processing provider may refuse a request.
This Instagram transcript generator transcribes speech, not the post description or text printed inside the video. Private, deleted, restricted, or sign-in-only posts cannot be unlocked here. Instagram can also refuse access to a public post.
Only English speech has been tested here. No support is claimed for other languages.
Recognition can miss or repeat words, particularly with music, overlapping voices, quiet speech and unfamiliar accents. Music-only and silent clips may return no text or words that were not spoken. Check every result against the audio. This tool does not translate.
GOOD TO KNOW
A few straight answers.
Is this an Instagram Reels to text tool?
Yes. It sends the public link through our cloud backend to a hosted speech service, which retrieves and transcribes the media. The post caption is not used as a substitute for a transcript.
Can I paste a Reel, post or IGTV link?
Yes. Direct Reel, single-video post and IGTV link formats are accepted. Profiles, Stories and multi-item collections are not accepted. An accepted link is not a promise that Instagram will make its audio available.
Do I need an account?
No account is needed for this tool. It does not sign in to Instagram or attempt to unlock content that requires an account.
Where do my link and transcript go?
Your link is sent to our Cloudflare backend and Supadata for video metadata and speech processing. Processing does not stay on your device. We temporarily cache completed text to avoid repeat charges and retain processing records for billing and failure review. There is no account or browsable user history; see the Privacy Policy for details.
Why does a public link sometimes fail?
A public post may still be inaccessible to automated retrieval. If Instagram returns a restriction or a network request fails, this page reports the error instead of inventing text. Try a different accessible public Reel.
Do I need to keep a local service running?
No. Use this website over HTTPS. Its cloud backend handles metadata checks and transcription; your computer does not need a local server or an installed speech model.