Google Cloud Speech-to-Text is a speech recognition API that turns audio into written text. It runs on Chirp 3, Google Cloud's universal speech model, and handles short clips, long recordings and live streaming audio sent from a microphone or Cloud Storage. Coverage spans 85+ languages and variants, with speaker diarization, multilingual language detection and model adaptation that biases results toward your own terms. New Google Cloud customers receive up to $300 in free credits to try it.
It is built for developers, ML engineers and product teams who need transcription inside their own software, from call analytics to media captioning. Enterprises in regulated sectors get API v2 features such as regionalized processing, audit logs in the Google Cloud console and customer-managed encryption keys. Non-coders can test audio files through a web interface in Agent Platform or the console demo before writing any code. Media and contact-center teams use it to caption archives and analyze recorded conversations at scale.
Common uses include transcribing recorded meetings and calls, captioning and subtitling video, adding voice input to mobile and web apps, and generating real-time captions for streaming content. A tutorial shows combining it with the Translation API to produce localized subtitles. Teams build searchable archives of interviews and lectures, index multi-speaker video, and feed transcripts to downstream analytics. Model adaptation helps when audio is noisy or full of product names and jargon that general models mishear. Developers can start from Cloud console quickstarts and sample applications before building a production pipeline, and results can be requested synchronously, asynchronously or as a stream.
