Skip to content

parakeet.md

Latest commit

 

History

History
115 lines (92 loc) · 5.52 KB

parakeet.md

File metadata and controls

115 lines (92 loc) · 5.52 KB
Raw

Parakeet

NVIDIA's Parakeet family ported to transcribe.cpp. A FastConformer encoder paired with one of three decoder heads — TDT (transducer with a duration prediction head), classical RNN-T, or CTC — and a TDT+CTC hybrid that ships both heads in one checkpoint. All variants take 16 kHz mono PCM through an 80-bin mel frontend; English-only across the family, with the single exception of parakeet-tdt-0.6b-v3 which extends to 25 European languages.

For the architecture deep-dive, validation contract, and porting notes, see the family doc at docs/porting/families/parakeet.md.

Choosing a variant

Most users want one of three:

  • English transcription → parakeet-tdt-0.6b-v2. The default pick: small, fast, near top-of-family accuracy on English.
  • Multilingual (25 European languages) → parakeet-tdt-0.6b-v3. The only multilingual variant; same size as v2, broader coverage at a small English-WER cost.
  • Streaming / real-time → Nemotron streaming. Parakeet here is offline-only. For low-latency streaming use the FastConformer-lineage nemotron-3.5-asr-streaming-0.6b (multilingual) or nemotron-speech-streaming-en-0.6b (English). Within Parakeet itself, parakeet-unified-en-0.6b is the only streaming-capable variant.

If you specifically need the lowest WER or a different decoder:

  • Lowest WER, English. parakeet-tdt-1.1b (1.38% Q8_0) is the most accurate in the family, narrowly ahead of parakeet-rnnt-1.1b (1.46%). TDT also decodes faster — the duration head lets the decoder skip frames.
  • Fastest decode at any size. Use a CTC variant (parakeet-ctc-0.6b / parakeet-ctc-1.1b). Single-pass greedy alignment, no transducer loop — at a ~0.2pp WER cost vs the same-size RNN-T.
  • Tiny footprint. parakeet-tdt_ctc-110m (135 MB at Q8_0) is the smallest Parakeet. The 1.1B tdt_ctc ships both heads but is primarily useful when you want TDT speed with CTC as a fallback at runtime.

All variants

WER is on LibriSpeech test-clean for the Q8_0 preset, measured by transcribe.cpp's WER pipeline. See each per-variant doc for the full quant matrix and the comparison to NVIDIA's self-reported numbers.

Variant Decoder Params Q8_0 size WER (Q8_0) Languages Doc
parakeet-tdt-0.6b-v2 TDT 0.6B 730 MB 1.69% English parakeet-tdt-0.6b-v2.md
parakeet-tdt-0.6b-v3 TDT 0.6B 740 MB 1.94% 25 European parakeet-tdt-0.6b-v3.md
parakeet-tdt-1.1b TDT 1.1B 1.27 GB 1.38% English parakeet-tdt-1.1b.md
parakeet-tdt_ctc-110m TDT+CTC 110M 135 MB 2.43% English parakeet-tdt_ctc-110m.md
parakeet-tdt_ctc-1.1b TDT+CTC 1.1B 1.27 GB 1.87% English parakeet-tdt_ctc-1.1b.md
parakeet-rnnt-0.6b RNN-T 0.6B 730 MB 1.62% English parakeet-rnnt-0.6b.md
parakeet-rnnt-1.1b RNN-T 1.1B 1.27 GB 1.46% English parakeet-rnnt-1.1b.md
parakeet-ctc-0.6b CTC 0.6B 722 MB 1.87% English parakeet-ctc-0.6b.md
parakeet-ctc-1.1b CTC 1.1B 1.26 GB 1.85% English parakeet-ctc-1.1b.md
parakeet-unified-en-0.6b RNN-T 0.6B 731 MB 1.60% English parakeet-unified-en-0.6b.md

Pre-built GGUFs for every variant and quant are hosted under handy-computer on Hugging Face; each per-variant doc has direct download links.

Input limits

No practical per-call length limit (transcribe_capabilities.max_audio_ms == 0): the Conformer encoder's positional encoding is recomputed per call, so audio of any length is processed in a single pass — pass arbitrarily long recordings. See the input-length contract.

Quick start

Pick a variant and run:

cmake -B build
cmake --build build

build/bin/transcribe-cli \
  -m models/parakeet-tdt-0.6b-v2/parakeet-tdt-0.6b-v2-Q8_0.gguf \
  samples/jfk.wav

The repo doesn't ship the GGUFs — pull them from the corresponding handy-computer/<variant>-gguf repo on Hugging Face, or convert from the upstream NVIDIA .nemo checkpoint via the per-variant doc's reproduction section.

Capabilities

All Parakeet variants support:

  • Transcription of 16 kHz mono WAV input.
  • Token-level timestamps at the encoder frame rate (TDT and RNN-T; CTC also exposes frame-level alignment).

Buffered streaming is supported on parakeet-unified-en-0.6b across all six published (L, C, R) configurations from the model's training menu (lookahead latency from 160ms at (70, 1, 1) through 2.08s at the default (70, 13, 13)). See parakeet-unified-en-0.6b.md for the per-config WER and the --stream-buf-{left,chunk,right}-ms CLI surface. Other Parakeet variants run offline only.

What's not supported (consistent across the family): translation, VAD, speaker diarization. Language coverage is English-only except parakeet-tdt-0.6b-v3 (25 European languages, no auto-detect — language hint required). See the family doc for the full runtime contract.