Transcribe Audio (Parakeet)

@nvidia1 VersionTranscribe

NVIDIA Parakeet-TDT (NeMo) transcriber. Resamples each AudioFrame to 16 kHz and emits oneof t[String, [Subtitle]] — plain text or timed segments. The v2 checkpoint is English-only and v3 covers a European language set.

How it fits

Typical backends

A recorded file is transcribed to timed subtitles and posted as JSON.

Only speech segments reach the transcriber, whose text drives an LLM follow-up.

Caveats

Hard constraintThe `model` must name a compatible NeMo ASR checkpoint; an unknown id fails to load at startup.
LimitationThe v2 checkpoint is English only and the v3 checkpoint covers a European multilingual set; there is no runtime language hint, so the model infers the language from the audio.
I/O contractThe plain-text arm returns one String; the subtitle arm returns per-segment start/end times. Empty input returns the empty result for the selected arm without running the model.
CostThe pretrained weights are a couple of gigabytes and the toolkit pulls a heavy dependency tree, so images are larger than the Whisper or Moonshine alternatives; pre-seed the model cache for offline deployments.
CompatibilityThe model is released under CC-BY-4.0; confirm compatibility with your deployment terms.
State lifetimeConfiguration is read once at startup; changing it requires a restart.

Versions

bd174c31latestdefault
Automated release
6/25/2026