Features
Word Search, Sentences, and Paragraphs
Find where words were spoken and fetch the transcript broken into timed sentences or paragraphs.
Three read endpoints slice a completed transcript into more useful shapes: search for specific words with their timestamps, or fetch the text segmented into sentences or paragraphs. All three require the transcript status to be completed (otherwise they return 409 with code not_ready).
Word search
/v2/transcripts/{transcript_id}/word-searchFind every occurrence of one or more words, with timestampsPass the terms in the q query parameter, comma-separated, up to 25 per request. Matching is case-insensitive and whole-word. Each match returns its count and the start and end timestamp (milliseconds) of every occurrence.
curl -s "https://api.realtimevoicekit.com/v2/transcripts/tr_9f2c31b8/word-search?q=pricing,renewal" \
-H "Authorization: Bearer rtvk_your_api_key"{
"query": "pricing, renewal",
"total_matches": 4,
"matches": [
{
"text": "pricing",
"count": 3,
"timestamps": [
{ "start": 12480, "end": 12920, "confidence": 0.99 },
{ "start": 84100, "end": 84590, "confidence": 0.98 },
{ "start": 132520, "end": 133060, "confidence": 0.99 }
]
},
{
"text": "renewal",
"count": 1,
"timestamps": [
{ "start": 48210, "end": 48800, "confidence": 0.97 }
]
}
]
}Sentences and paragraphs
/v2/transcripts/{transcript_id}/sentencesThe transcript split into timed sentences/v2/transcripts/{transcript_id}/paragraphsThe transcript grouped into timed paragraphsBoth endpoints return the same segment shape: an ordered data array where each entry has an index, the segment text, start and end timestamps in milliseconds, an average confidence, and the underlying words.
{
"data": [
{
"index": 0,
"text": "Welcome back to the show.",
"start": 130,
"end": 1980,
"confidence": 0.98,
"words": [
{ "text": "Welcome", "start": 130, "end": 610, "confidence": 0.99 }
]
}
]
}