Palatine SpeechInstall

Summary

Convert speech in an uploaded binary audio file to text using Palatine Speech. Submit the file with the selected model and automatically poll for completion when the service returns a task ID.

Inputs

• model — Choose palatine_small (default), described as fast and compact, or palatine_large_highspeed, described as slower and more accurate.
• binaryProperty (required) — Name of the input binary property containing the audio file; defaults to data. The upload helper reads file metadata from the first input item.
• polling — Optional polling interval and attempt limit, defaulting to 500 ms and 200 attempts. Runtime clamps these to 500–900000 ms and 1–1000 attempts; values of 1000 ms or 300 attempts are reset to the defaults.
• requestOptions — Declares optional batching, SSL certificate validation bypass, proxy, and request timeout settings. These settings are absent from the node's source properties and are not explicitly consumed by its helpers; polling requests do not forward them.

Output shape

JSON items containing task: "transcribe", source_file, transcription, and model_used when a model is set, with the original binary data preserved.

The output hook maps the same response to each item it receives. Transcription uses returned text or transcription fields, otherwise serializes the payload into a string; it does not expose raw task metadata or timestamps. source_file is empty when no filename is available. Polling throws an error if the task fails or exceeds the attempt limit.

Links

• n8n community nodes installation guide