parakeet-ctc-1.1b is an ASR model that transcribes speech in lower case English alphabet. This model is jointly developed by NVIDIA NeMo and Suno.ai teams. It is an XXL version of FastConformer CTC [1] (around 1.1B parameters) model.
Parameters1.1B
Context—
Weights9.7 GB
Licensecc-by-4.0
AccessOpen weights
Monthly Downloads796.2k
Runs On
What it takes to serve parakeet-ctc-1.1b (1.1B parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 2.1 GB | 2.6 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 1.1 GB | 1.3 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 0.5 GB | 0.6 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Sep 18, 2026.
SAVRN's Notes on parakeet-ctc-1.1b
We would give this one the smallest slice of a card we could carve out. NVIDIA's 1.1B-parameter speech recognizer, built with Suno.ai on FastConformer CTC and served through NeMo, needs 2.6 GB at 16-bit, 1.3 GB at 8-bit and 0.6 GB at 4-bit, and it writes lower-case English. On the cheapest host we price, one MI300X with 192 GB at $1.85 an hour on-demand, that footprint is a rounding error, so transcription belongs beside everything else on the card, not on a second one.
CC BY 4.0 lets you share and adapt it, commercially included, as long as NVIDIA is credited and your changes are indicated. Before committing, look at the training mix, librispeech_asr, fisher_corpus, Switchboard-1, WSJ and vctk among them, and at NVIDIA's own reported word error rates, 1.83 on LibriSpeech clean up to 15.62 on AMI meetings, then run your own audio through it. Released December 28, 2023, last updated August 5, 2026.
Model Card
By NVIDIA, published under cc-by-4.0, revision 20e63a0fed6a.
Parakeet CTC 1.1B (en)
| |
parakeet-ctc-1.1b is an ASR model that transcribes speech in lower case English alphabet. This model is jointly developed by NVIDIA NeMo and Suno.ai teams.
It is an XXL version of FastConformer CTC [1] (around 1.1B parameters) model.
See the model architecture section and NeMo documentation for complete architecture details.
NVIDIA NeMo: Training
To train, fine-tune or play with the model you will need to install NVIDIA NeMo. We recommend you install it after you've installed latest PyTorch version.
pip install nemo_toolkit['all']
How to Use this Model
There are several ways to use this model. Choose the one that fits your needs.
Run locally with NeMo-Speech.cpp
NeMo-Speech.cpp provides a lightweight native C++ runtime for local inference with this model. After installing the runtime:
hf download nvidia/parakeet-ctc-1.1b \
parakeet-ctc-1.1b.q8_0.gguf \
--local-dir models
nemo-speech transcribe audio.wav \
--model models/parakeet-ctc-1.1b.q8_0.gguf
See the NeMo-Speech.cpp documentation for more details.
NVIDIA NeMo
Configuration
- Architecture
- ParakeetForCTC
- Vocabulary size
- 1,025
- Model type
- parakeet_ctc
Identity and Version
- Repository
- nvidia/parakeet-ctc-1.1b
- Publisher
- NVIDIA
- Task
- Speech recognition
- Modality
- Audio
- Library
- nemo
- Parameters
- 1.1B parameters
- Languages
- en
- Revision
- 20e63a0fed6aedba145b74b826dbd41df0941730
- First published
- 2023-12-28
- Last updated
- 2026-08-05
Files and Weights
11 files, 9.7 GB in total. The weights are 3 files totalling 9.7 GB in gguf, nemo, safetensors.
Weights3 files · 9.7 GB
Configuration3 files · 1.6 KB
Tokenizer3 files · 429.5 KB
Documentation1 file · 13.0 KB
Repository1 file · 1.7 KB
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model.safetensors | Weights | 4.3 GB | 57e0bc26772f |
| parakeet-ctc-1.1b.nemo | Weights | 4.3 GB | 8e91253dd138 |
| parakeet-ctc-1.1b.q8_0.gguf | Weights | 1.2 GB | 6584fc0fdacf |
| config.json | Configuration | 962 B | — |
| preprocessor_config.json | Configuration | 314 B | — |
| special_tokens_map.json | Configuration | 279 B | — |
| README.md | Documentation | 13.0 KB | — |
| .gitattributes | Repository | 1.7 KB | — |
| tokenizer.json | Tokenizer | 412.4 KB | — |
| tokenizer_config.json | Tokenizer | 634 B | — |
| vocab.json | Tokenizer | 16.5 KB | — |
License and Download
- License
- cc-by-4.0
- Access
- Open weights, no gate
- Download size
- 9.7 GB
Download from NVIDIA
Released by NVIDIA through its official repository on Hugging Face. Read the license.
Built From
- Described by arXiv:2305.05084
- Trained on (disclosed) MLCommons/peoples_speech
- Trained on (disclosed) National-Singapore-Corpus-Part-1
- Trained on (disclosed) National-Singapore-Corpus-Part-6
- Trained on (disclosed) Switchboard-1
- Trained on (disclosed) WSJ-0
- Trained on (disclosed) WSJ-1
- Trained on (disclosed) europarl
- Trained on (disclosed) fisher_corpus
- Trained on (disclosed) librispeech_asr
- Trained on (disclosed) mozilla-foundation/common_voice_8_0
- Trained on (disclosed) multilingual_librispeech
- Trained on (disclosed) vctk
- Trained on (disclosed) voxpopuli
Evaluations
Each result is shown as reported, with the conditions its reporter stated. None is a SAVRN measurement. A comparison lines two results up only when their configuration, unit and setup are all stated and identical.
| Benchmark | Conditions | Result | Reported by | Revision | Date |
|---|---|---|---|---|---|
| AMI (Meetings test) | Configuration ihmTask Automatic Speech RecognitionMetric Test WERComparison conditions not established | 15.62 | nvidia Publisher reported |
Evaluated revision not stated | — |
| Earnings-22 | Task Automatic Speech RecognitionMetric Test WERComparison conditions not established | 13.69 | nvidia Publisher reported |
Evaluated revision not stated | — |
| GigaSpeech | Task Automatic Speech RecognitionMetric Test WERComparison conditions not established | 10.27 | nvidia Publisher reported |
Evaluated revision not stated | — |
| LibriSpeech (clean) | Configuration otherTask Automatic Speech RecognitionMetric Test WERComparison conditions not established | 1.83 | nvidia Publisher reported |
Evaluated revision not stated | — |
| LibriSpeech (other) | Configuration otherTask Automatic Speech RecognitionMetric Test WERComparison conditions not established | 3.54 | nvidia Publisher reported |
Evaluated revision not stated | — |
| Mozilla Common Voice 9.0 | Configuration enTask automatic-speech-recognitionMetric Test WERComparison conditions not established | 9.02 | nvidia Publisher reported |
Evaluated revision not stated | — |
| SPGI Speech | Configuration testTask automatic-speech-recognitionMetric Test WERComparison conditions not established | 4.2 | nvidia Publisher reported |
Evaluated revision not stated | — |
| Vox Populi | Configuration enTask Automatic Speech RecognitionMetric Test WERComparison conditions not established | 6.53 | nvidia Publisher reported |
Evaluated revision not stated | — |
| hf-audio/open-asr-leaderboard | Task ami_werMetric ami_werComparison conditions not established | 15.67 | open-asr-leaderboard Reported by a third party |
Evaluated revision not stated | 2023-12-28 |
| hf-audio/open-asr-leaderboard | Task earnings22_werMetric earnings22_werComparison conditions not established | 13.75 | open-asr-leaderboard Reported by a third party |
Evaluated revision not stated | 2023-12-28 |
| hf-audio/open-asr-leaderboard | Task gigaspeech_werMetric gigaspeech_werComparison conditions not established | 10.28 | open-asr-leaderboard Reported by a third party |
Evaluated revision not stated | 2023-12-28 |
| hf-audio/open-asr-leaderboard | Task librispeech_clean_werMetric librispeech_clean_werComparison conditions not established | 1.83 | open-asr-leaderboard Reported by a third party |
Evaluated revision not stated | 2023-12-28 |
| hf-audio/open-asr-leaderboard | Task librispeech_other_werMetric librispeech_other_werComparison conditions not established | 3.51 | open-asr-leaderboard Reported by a third party |
Evaluated revision not stated | 2023-12-28 |
| hf-audio/open-asr-leaderboard | Task mean_werMetric mean_werComparison conditions not established | 7.39875 | open-asr-leaderboard Reported by a third party |
Evaluated revision not stated | 2023-12-28 |
| hf-audio/open-asr-leaderboard | Task rtfxMetric rtfxComparison conditions not established | 2728.52 | open-asr-leaderboard Reported by a third party |
Evaluated revision not stated | 2023-12-28 |
| hf-audio/open-asr-leaderboard | Task spgispeech_werMetric spgispeech_werComparison conditions not established | 4.02 | open-asr-leaderboard Reported by a third party |
Evaluated revision not stated | 2023-12-28 |
| hf-audio/open-asr-leaderboard | Task tedlium_werMetric tedlium_werComparison conditions not established | 3.57 | open-asr-leaderboard Reported by a third party |
Evaluated revision not stated | 2023-12-28 |
| hf-audio/open-asr-leaderboard | Task voxpopuli_werMetric voxpopuli_werComparison conditions not established | 6.56 | open-asr-leaderboard Reported by a third party |
Evaluated revision not stated | 2023-12-28 |
| tedlium-v3 | Configuration release1Task automatic-speech-recognitionMetric Test WERComparison conditions not established | 3.54 | nvidia Publisher reported |
Evaluated revision not stated | — |
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 9.7 GB |
| 16-bit | 2.1 GB |
| 8-bit | 1.1 GB |
| 4-bit | 0.5 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Compare parakeet-ctc-1.1b
Questions About parakeet-ctc-1.1b
How much GPU memory does parakeet-ctc-1.1b need?
About 2.6 GB at 16-bit and 0.6 GB at 4-bit: the weights (1.1B parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run parakeet-ctc-1.1b on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Can I use parakeet-ctc-1.1b commercially?
Yes. parakeet-ctc-1.1b is released under Creative Commons Attribution 4.0. CC BY 4.0 permits sharing and adapting the work, including commercially, provided the creator is credited and changes are indicated.
Similar Models
This model is finetuned on top of feature extractor XLS-R from Facebook/Meta. The finetuned model achieves the following results on the test set with a 5-gram KenLM. The numbers in parentheses are the results without the language model: This is one of several Wav2Vec-models our team created during the hosted Robust Speech Event. This is the complete list of our models and their final scores: In parallel with the event, the team also converted the Norwegian Parliamentary Speech Corpus (NPSC) to the NbAiLab/NPSC in Dataset format and used that as the main source for training. We have released all the code developed during the event so that the Norwegian NLP community can build upon it when…
Whisper is a state-of-the-art model for automatic speech recognition (ASR) and speech translation, proposed in the paper Robust Speech Recognition via Large-Scale Weak Supervision by Alec Radford et al. from OpenAI. Trained on >5M hours of labeled data, Whisper demonstrates a strong ability to generalise to many datasets and domains in a zero-shot setting. Whisper large-v3-turbo is a finetuned version of a pruned Whisper large-v3. In other words, it's the exact same model, except that the number of decoding layers have reduced from 32 to 4. As a result, the model is way faster, at the expense of a minor quality degradation. You can find more details about it in this GitHub discussion.…
Fine-tuned openai/whisper-large-v3-turbo on Monsoon Hindi ASR data using LoRA (r=32, alpha=64). English capabilities preserved with <10% relative degradation across all benchmarks. LoRA's structural separation (frozen base + low-rank adapters) provides a natural safeguard against catastrophic forgetting.
For usage instructions follow openai/whisper-large-v3-turbo. Note for faster-whisper vocab changes make model.ismultilingual and suppresstokens wrong. Please adjust the code as required if you want to use this with faster-whisper. Turbo finetune with japanese tokenizer. Full finetune trained 2^19 steps, batch size 64. Smaller vocab with ~1.6x bytes/token allows faster speed with 4 layers vs 2 layer distil (10% larger decoder). Benchmarks. Short form slightly behind v0.2 (trained less?) but long form much better. Also trained for lyrics but untested. Research supported with Cloud TPUs from Google's TPU Research Cloud (TRC)
Whisper finetune for Japanese focused on general/anime domains. For usage instructions follow openai/whisper-large-v3-turbo. Due to vocab changes ctranslate2>=4.7.1 required for faster-whisper. For inference engines with hardcoded vocab, the token embedding can be padded. Finetuned from turbo with pruned vocab, indices can be found in mapping.txt. Trained decoder only for 2^20 steps, batch size 64. Using a 45000 hour corpus (largest source is 17000 of filtered reazonspeech-all) with custom mixing ratio and augmentation to maintain long form performance and timestamps. Benchmarks. Competitive/SOTA on test sets, slightly better than 1.5B on short form, worse on long form. Also trained for…