Parakeet (NeMo Speech)
Parakeet is NVIDIA's family of speech recognition models, run through the NeMo Speech toolkit. The toolkit code is Apache-2.0 and the Parakeet TDT 0.6B v3 weights are CC-BY-4.0, which permits commercial use with attribution. The model covers 25 European languages with automatic detection, punctuation, word timestamps and long audio. It expects an NVIDIA GPU, which is why desktop apps bundle it for CPU use instead.
What does Parakeet (NeMo Speech) do?
NeMo Speech is NVIDIA's toolkit "for researchers and PyTorch developers working on Speech models", covering speech recognition, text to speech and speech language models. The model most people mean when they say Parakeet is parakeet-tdt-0.6b-v3: "a 600-million-parameter multilingual automatic speech recognition (ASR) model designed for high-throughput speech-to-text transcription", covering 25 European languages and detecting the language automatically.
The model card lists automatic punctuation and capitalisation, word-level and segment-level timestamps, and long audio: up to 24 minutes with full attention on an A100 80 GB, or up to 3 hours with local attention. The README also names newer models: Nemotron ASR Streaming with 40 languages and controllable latency from 80 milliseconds to 1 second, a unified English model that does offline and streaming inference in one network, and Canary for recognition plus translation.
Key facts
- Licence: Apache-2.0 on the NeMo Speech toolkit; CC-BY-4.0 on the parakeet-tdt-0.6b-v3 weights, stated on the Hugging Face model card.
- Parakeet TDT 0.6B v3 is a 600-million-parameter model "designed for high-throughput speech-to-text transcription", per its card.
- 25 European languages, with the language detected automatically.
- Automatic punctuation and capitalisation, and word-level and segment-level timestamps.
- Long audio: up to 24 minutes with full attention on an A100 80 GB, or up to 3 hours with local attention.
- Newer models named in the README: Nemotron ASR Streaming with 40 languages and latency from 80 milliseconds to 1 second; a unified English model doing offline and streaming in one network; Canary for recognition plus translation.
- Install: pip install 'nemo-toolkit[asr,tts]', a source install with uv, or the NGC Docker container; Python 3.12 and PyTorch 2.7 or later.
- The card lists at least 2 GB of RAM to load the model and names NVIDIA Ampere, Blackwell, Hopper and Volta as supported; a Transformers path with a CPU fallback is documented.
What does it replace, and where does it fall short?
Parakeet replaces Deepgram and AssemblyAI for high-volume transcription of European languages on your own NVIDIA hardware, with the timestamps and punctuation those APIs return.
The label is open weights rather than open source because the weights are CC-BY-4.0, a Creative Commons licence that is not on the OSI list, and the card says training combined a public corpus with an in-house NVIDIA dataset. CC-BY-4.0 does permit commercial use, with attribution. The code around it is Apache-2.0.
Where it falls short: the toolkit expects an NVIDIA GPU. Speaker labelling and quantisation are not described in the README. The desktop dictation apps on this site (Handy, VoiceInk, OpenWhispr, Vibe) already bundle Parakeet for CPU use, which is the easier route for a single machine.
How does Parakeet (NeMo Speech) run?
pip install 'nemo-toolkit[asr,tts]', a source install with uv, or the NGC Docker container run with all GPUs attached. Python 3.12 or later and PyTorch 2.7 or later. The README states "NVIDIA GPU + CUDA (required for training; recommended for inference)". The model card documents a Transformers path with a CPU fallback and a NeMo-Speech.cpp runtime. NVIDIA links a hosted NIM for the streaming model; no prices are stated.
Who is Parakeet (NeMo Speech) for?
A team with NVIDIA GPUs transcribing many hours of European-language audio a day. A team on Macs or CPUs should use whisper.cpp, or one of the desktop apps that already ships Parakeet.
What limits does the README state?
Some checkpoints need loading with weights_only off, which the README warns "can have the risk of arbitrary code execution" from untrusted files. The card: "Transcripts may be not 100% accurate", and words absent from the vocabulary "not likely to be recognized".
Questions people ask
Is Parakeet open source?
The toolkit is, under Apache-2.0. The weights are CC-BY-4.0, a Creative Commons licence that permits commercial use with attribution but is not on the OSI list, and the card says training combined a public corpus with an in-house NVIDIA dataset. That is why this site labels Parakeet open weights.
Can Parakeet be used commercially?
Yes. CC-BY-4.0 requires attribution and nothing else, and the model card calls it a permissive licence. The Apache-2.0 toolkit around it permits commercial use too. Several desktop apps on this site, including Handy and OpenWhispr, ship Parakeet models.
Does Parakeet need an NVIDIA GPU?
The toolkit says an NVIDIA GPU with CUDA is required for training and recommended for inference. The model card documents a Hugging Face Transformers path with a CPU fallback and a NeMo-Speech.cpp runtime. For a single laptop the practical route is a desktop app that already bundles Parakeet for CPU use.
How does Parakeet compare with Deepgram?
Parakeet transcribes 25 European languages at volume on your own NVIDIA hardware with timestamps and punctuation, and no per-minute bill. Deepgram is a hosted API with streaming and conversation features. The alternatives to Deepgram page places Parakeet as the pick for high volume in European languages on NVIDIA.
Which languages does Parakeet TDT v3 cover?
Its card says 25 European languages with automatic language detection. The README also names Nemotron ASR Streaming with 40 languages and Canary with 25 European languages plus translation. For languages outside Europe, Whisper and Moonshine are the options on this site.
Sources
- NeMo Speech README and Apache-2.0 LICENSE: github.com/NVIDIA-NeMo/Speech, read 2026-09-16.
- Parakeet TDT 0.6B v3 model card, CC-BY-4.0: huggingface.co/nvidia/parakeet-tdt-0.6b-v3, read 2026-09-16.
Compared with the others
On the open-source alternatives to Deepgram page, Parakeet (NeMo Speech) is the pick for high volume in european languages on nvidia. Apache-2.0 code and CC-BY-4.0 weights. Parakeet TDT 0.6B v3 covers 25 European languages with automatic detection, punctuation, word timestamps and long audio.
Also on that page: Whisper for the reference model on a gpu, whisper.cpp for a cpu, a mac, or a phone, WhisperX for word timestamps and speaker labels.
On the open-source alternatives to AssemblyAI page, Parakeet (NeMo Speech) is the pick for high volume in european languages on nvidia. Apache-2.0 code and CC-BY-4.0 weights. Parakeet TDT 0.6B v3 covers 25 European languages with automatic detection, punctuation, word timestamps and long audio.
Also on that page: WhisperX for word timestamps and speaker labels, Speaches for an openai-compatible server, Whisper for the reference model on a gpu.
More speech-to-text engines
OpenAI's general-purpose speech recognition model, released with code and weights under MIT, in six sizes from 39 million to 1.55 billion parameters, with multilingual transcription, translation to English and language detection.
A dependency-free C and C++ port of Whisper under MIT that runs on CPUs, Apple Silicon and phones, with quantised models, an HTTP server and bindings for a dozen languages.
Batched Whisper transcription under BSD-2-Clause with word-level timestamps from forced alignment and speaker labels from pyannote, the closest open-source match to a hosted transcription API's output.
A runtime under Apache-2.0 from the next-generation Kaldi team that runs speech to text, text to speech, speaker diarization and voice activity detection locally on CPUs, NPUs and phones, with no internet connection.
An offline speech recognition toolkit under Apache-2.0 with streaming, models of about 50 MB, 20+ languages, a reconfigurable vocabulary and speaker identification, from small boards to clusters.
An on-device speech toolkit under MIT built for live streaming, with speech-to-text models trained from scratch in sizes down to 1 MB, one library across Python, the browser, phones and desktops.
Added September 16, 2026. Every claim above comes from the project's README, LICENSE or model card, read on September 16, 2026, or from the GitHub API on the date shown in the panel. Found an error? Write to reveneau@licheo.com and it is fixed in the next weekly pass. Repository: github.com/NVIDIA-NeMo/Speech.
Want Parakeet (NeMo Speech) running on your own servers?
Reveneau sets it up on your hardware, connects it to your product, and builds the software around it.