Instagram Reel transcript API
Transcribe any public Instagram reel or video post: the full speech transcript, speaker labels, and word-level timestamps, from a reel, /p/, or /tv/ URL or an Instagram CDN media URL you already hold. A video post is transcribed exactly like a reel. A photo post comes back found with its record and an empty transcript, charged the request price only. Transcription runs on MAI-Transcribe-2, chosen for its accuracy and its speaker labels. Turn on hostVideo to also get the MP4 on a hosted link that plays without an Instagram session. If you only want the text, instagram.media_transcript is the cheaper transcript-only option; if you only want the file, instagram.post is where you go.
Try it
Make your first request
{
"data": {
"items": [
{
"bytes": 42,
"durationSeconds": 12.5,
"expiresUtc": 12.5,
"hostedUrl": "https://example.com/page",
"id": "a1b2c3d4",
"language": "en",
"likeCount": 12500,
"mediaType": "general",
"ownerUsername": "alex_rivera",
"segments": [
{
"end": 12.5,
"speaker": "example",
"start": 12.5,
"text": "A short example description of this item.",
"words": [
null
]
}
],
"shortcode": "a1b2c3d4",
"text": "A short example description of this item.",
"thumbnailUrl": "https://example.com/image.jpg",
"url": "https://example.com/page"
}
]
},
"found": true
}interface InstagramReelTranscriptResponse {
data: {
items: {
bytes?: number;
durationSeconds?: number;
expiresUtc?: number;
hostedUrl?: string;
id: string;
language?: string;
likeCount?: number;
mediaType?: string;
ownerUsername?: string;
segments?: {
end?: number;
speaker?: string;
start?: number;
text?: string;
words?: {
end?: number;
start?: number;
text?: string;
}[];
}[];
shortcode?: string;
text: string;
thumbnailUrl?: string;
url: string;
}[];
} | null;
found: boolean;
}Full parameter and response reference - every field, type, and example for this endpoint.
Reference
Request, response, and price
Last verified 2026-09-17 · uptime and latency measured over 30dcurl -X POST https://api.getanyapi.com/v1/run/instagram.reel_transcript \
-H "Authorization: Bearer $ANYAPI_KEY" \
-H "Content-Type: application/json" \
-d '{"hostVideo":true,"url":"https://www.instagram.com/reel/CfY6jCIgH-P/","wordTimestamps":false}'| Field | Type | Example value |
|---|---|---|
| Request body | ||
| url | string | "https://www.instagram.com/reel/CfY6jCIgH-P/"The URL of a public Instagram reel or video post (e.g. https://www.instagram.com/reel/C8yKXdRxKqK/), or an Instagram CDN media URL you already hold. |
| hostVideo | boolean | trueSet true to also get the reel's MP4 on a hosted link you can play without an Instagram session. Charged as an extra on top of the transcript (e.g. true). A photo post has no file to host, so a request that sets this on one is refused with no charge; send it without hostVideo to get the post record instead. |
| wordTimestamps | boolean | falseSet true to include a precise timestamp for every word in the transcript (e.g. true). |
| Response | ||
| data | object | |
| data.items | object[] | Record for the requested reel (one item), with the full transcript text, timed segments, source video metadata, and the hosted video link when hostVideo was asked for. Populated whenever the provider has data for the entity. |
| data.items[].bytes | integer | Size of the hosted MP4 in bytes. Present only when hostVideo was true. |
| data.items[].durationSeconds | number | Video duration in seconds. Absent when there is no transcript, which is what measures it. |
| data.items[].expiresUtc | number | UTC epoch timestamp in seconds (Unix time). Multiply by 1000 for a JS Date in milliseconds. After this moment the hosted MP4 is deleted. Present only when hostVideo was true. |
| data.items[].hostedUrl | string | A direct link to the downloaded MP4, hosted by AnyAPI and playable without an Instagram session. Present only when hostVideo was true. |
| data.items[].id | string | The reel's numeric Instagram media ID, as a string. Empty when the request supplied a CDN media URL, which carries no post record. Populated whenever the provider has data for the entity. |
| data.items[].language | string | Detected spoken language (ISO 639-1 code, e.g. "en"). Absent when there is no transcript. |
| data.items[].likeCount | integer | Number of likes on the reel. Absent when the lane that served the lookup does not carry a like count. |
| data.items[].mediaType | string | What kind of media this is, as Instagram labels it (for example a video or an image post). On a photo post this is how you tell that the empty transcript is the answer rather than silence. |
| data.items[].ownerUsername | string | Username of the reel's owner, without the @ prefix. Empty when the request supplied a CDN media URL. |
| data.items[].segments | object[] | Time-aligned transcript segments in playback order, one per sentence, each with its text, speaker label, and start/end offsets in seconds. A segment starts when its first word is spoken and ends when its last one finishes. Empty when the reel has no detectable spoken audio. |
| data.items[].segments[].end | number | Segment end offset in seconds from the start of the video. |
| data.items[].segments[].speaker | string | Which speaker said this segment, as a stable label within this transcript (e.g. "0", "1"). |
| data.items[].segments[].start | number | Segment start offset in seconds from the start of the video. |
| data.items[].segments[].text | string | The segment's transcribed text. |
| data.items[].segments[].words | object[] | Every word in the segment with its own timing. Present only when wordTimestamps was true. |
| data.items[].segments[].words[].end | number | Word end offset in seconds from the start of the video. |
| data.items[].segments[].words[].start | number | Word start offset in seconds from the start of the video. |
| data.items[].segments[].words[].text | string | The word as spoken. |
| data.items[].shortcode | string | The reel's short code, the part of its instagram.com URL after /reel/. Empty when the request supplied a CDN media URL. |
| data.items[].text | string | The full speech transcript. Empty when the reel has no detectable spoken audio, or when the post is a photo with no video to transcribe. Populated whenever the provider has data for the entity. |
| data.items[].thumbnailUrl | string | A link to the reel's cover image on Instagram. This link is signed by Instagram and stops working after a short time. |
| data.items[].url | string | The reel URL the request asked for, returned as sent. Populated whenever the provider has data for the entity. |
| found | boolean | Whether the post was found. A reel or video post is found and transcribed. A photo post is found too: its record comes back with an empty transcript and is charged the request price only, with no audio minutes. A post that does not exist is not found and is free. |
| Price | ||
| Price per request (ceiling) | USD | $0.095 |
| Price /1k audio minutes | USD | $5.94 |
FAQ
About the Instagram Reel transcript API
The AnyAPI Instagram Reel transcript API returns Instagram reel transcript data as normalized JSON from one POST call to /v1/run/instagram.reel_transcript. Transcribe any public Instagram reel or video post: the full speech transcript, speaker labels, and word-level timestamps, from a reel, /p/, or /tv/ URL or an Instagram CDN media URL you already hold. A video post is transcribed exactly like a reel. A photo post comes back found with its record and an empty transcript, charged the request price only. Transcription runs on MAI-Transcribe-2, chosen for its accuracy and its speaker labels. Turn on hostVideo to also get the MP4 on a hosted link that plays without an Instagram session. If you only want the text, instagram.media_transcript is the cheaper transcript-only option; if you only want the file, instagram.post is where you go. AnyAPI returns one normalized schema whichever source serves it. It costs $5.94 per 1,000 audio minutes, and never more than $0.095 for a single request, in US dollars with no subscription and no monthly minimum. Over the last 30 days, 99.0% of Instagram reel transcript calls through AnyAPI succeeded, with a median response time of 32.1 seconds across 101 measured calls.