Google Cloud Speech

Convert audio to text transcriptions and synthesize natural-sounding speech from text using Google's neural network models. Perform synchronous, asynchronous, and streaming speech-to-text recognition across 125+ languages. Create and manage recognizer configurations for reusable transcription settings. Adapt speech models with custom phrase sets, custom classes, and boost values to improve accuracy for domain-specific vocabulary. Identify distinct speakers via speaker diarization and recognize multi-channel audio. Generate subtitle/caption output in SRT format. Synthesize text or SSML into audio using Standard, WaveNet, Neural2, Studio, and Chirp voice types with configurable pitch, speaking rate, volume, and encoding. Produce long-form audio content asynchronously.

Tools

Batch Transcribe Audio

Start an asynchronous batch transcription of one or more audio files stored in Google Cloud Storage. Returns a long-running operation that can be monitored using the Get Operation tool. Suitable for audio files longer than 1 minute (up to 8 hours). Results can be written to a GCS output location or returned inline when the operation completes.

Get Operation

Check the status and retrieve results of a long-running Speech-to-Text operation. Use this to monitor batch transcription jobs started with the Batch Transcribe Audio tool. Returns the current status, and when complete, the full transcription results or error details.

List Voices

List available Text-to-Speech voices. Optionally filter by language code to find voices for a specific language. Returns voice names, genders, supported languages, and native sample rates.

Create Recognizer

Create a named recognizer configuration for Speech-to-Text v2. A recognizer stores default settings like model, language, and recognition features so they don't need to be repeated in every transcription request.

Synthesize Speech

Convert text or SSML into natural-sounding speech audio using Google Cloud Text-to-Speech. Returns base64-encoded audio data in the requested format. Supports multiple voice types including Standard, WaveNet, Neural2, Studio, and Chirp 3 HD voices. Customize pitch, speaking rate, and volume.

Transcribe Audio

Transcribe audio to text using Google Cloud Speech-to-Text (synchronous recognition). Supports inline base64-encoded audio or audio files in Google Cloud Storage. Use for audio files up to 1 minute in duration. Configure language, model, punctuation, word-level details, speaker diarization, and speech adaptation hints.

Notes

• For Speech-to-Text v2 recognizers and inline transcription, global is the safest location default. Google documents v2 resource paths as projects/{project}/locations/{location}, but the recognizer examples consistently use locations/global, and some projects return API errors if a regional location is used for these flows.

License

This integration is licensed under the FSL-1.1.