API de X / Twitter: Transcripción de tweet
Extrae la transcripción hablada de un tweet con video de Twitter/X por URL.
US$1.20/1k solic., pagas por solicitud, sin suscripción.
Post
buildwithjonah
@buildwithjonah
Transcript
Alright so I spent the last six months building a data pipeline that normalizes about forty different scraper APIs into one schema. I want to walk through what broke and what I would do differently if I started over today. The first lesson was that every provider's error format is completely different. Some return a 200 with an error field buried three levels deep, some return a real HTTP error code, and some just silently return an empty array. We ended up writing a translation layer just so our retry logic actually knew whether a failure meant retry later or never retry at all. The second lesson was around pagination. Cursors are never portable across providers, so we had to build our own cursor abstraction just to keep the client code simple and predictable.
{"transcript": "Alright so I spent the last six months building a data pipeline that normalizes about forty different scraper APIs into one schema. I want to walk through what broke and what I would do differently if I started over today. The first lesson was that every provider's error format is completely different. Some return a 200 with an error field buried three levels deep, some return a real HTTP error code, and some just silently return an empty array. We ended up writing a translation layer just so our retry logic actually knew whether a failure meant retry later or never retry at all. The second lesson was around pagination. Cursors are never portable across providers, so we had to build our own cursor abstraction just to keep the client code simple and predictable."}
Precios y proveedores
Precios del endpoint Transcripción de tweet de la API de X / Twitter: una API, 2 proveedores intercambiables
El proveedor más barato responde primero. Si falla, el siguiente toma el relevo en la misma llamada.
Llámalo
Obtén Transcripción de tweet de X / Twitter en una sola solicitud
Cargando...
Referencia completa de parámetros y respuesta: cada campo, tipo y ejemplo de este endpoint.
Referencia
Una solicitud, un formato de respuesta.
Cada campo que envías y cada campo que recibes, cada uno con un ejemplo real.
Verificado por última vez el 2026-10-06 · disponibilidad y latencia medidas en 30d
Cuerpo de la solicitud
JSON, enviado por POST a este endpoint.
- urlstring
https://x.com/buildwithjonah/status/1842093771280384512
Respuesta
JSON, un valor de ejemplo por campo.
- foundboolean
true
- transcriptstring
Alright so I spent the last six months building a data pipeline that normalizes about forty different scraper APIs into one schema. I want to walk through what broke and what I would do differently if I started over today. The first lesson was that every provider's error format is completely different. Some return a 200 with an error field buried three levels deep, some return a real HTTP error code, and some just silently return an empty array. We ended up writing a translation layer just so our retry logic actually knew whether a failure meant retry later or never retry at all. The second lesson was around pagination. Cursors are never portable across providers, so we had to build our own cursor abstraction just to keep the client code simple and predictable.
Preguntas frecuentes
Acerca del endpoint Transcripción de tweet de la API de X / Twitter
El endpoint Transcripción de tweet de AnyAPI para X / Twitter devuelve datos de X / Twitter como JSON normalizado con una sola llamada POST a /v1/run/twitter.tweet_transcript. Extrae la transcripción hablada de un tweet con video de Twitter/X por URL. AnyAPI enruta cada solicitud entre 2 fuentes y recurre automáticamente a otra cuando una falla. Cuesta desde US$1.20 por cada 1,000 solicitudes, en dólares estadounidenses, sin suscripción y sin mínimo mensual. En los últimos 30 días, el 100.0% de las llamadas al endpoint Transcripción de tweet de la API de X / Twitter a través de AnyAPI tuvo éxito, con un tiempo de respuesta mediano de 3.6 segundos en 20 llamadas medidas.