parakeet-tdt-0.6b-v3 is a 600-million-parameter multilingual automatic speech recognition (ASR) model designed for high-throughput speech-to-text transcription. It extends the parakeet-tdt-0.6b-v2 model by expanding language support from English to 25 European languages. The model automatically detects the language of the audio and transcribes it without requiring additional prompting. It is part of a series of models that leverage the Granary [1, 2] multilingual corpus as their primary training dataset.
Supported Languages:
Bulgarian (bg), Croatian (hr), Czech (cs), Danish (da), Dutch (nl), English (en), Estonian (et), Finnish (fi), French (fr), German (de), Greek (el), Hungarian (hu), Italian (it), Latvian (lv), Lithuanian (lt), Maltese (mt), Polish (pl), Portuguese (pt), Romanian (ro), Slovak (sk), Slovenian (sl), Spanish (es), Swedish (sv), Russian (ru), Ukrainian (uk)
This model is ready for commercial/non-commercial use.
Key Features:
parakeet-tdt-0.6b-v3's key features are built on the foundation of its predecessor, parakeet-tdt-0.6b-v2, and include:
Automatic punctuation and capitalization
Accurate word-level and segment-level timestamps
Long audio transcription, supporting audio up to 24 minutes long with full attention (on A100 80GB) or up to 3 hours with local attention.
Released under a permissive CC BY 4.0 license
For full details on the model architecture, training methodology, datasets, and evaluation results, check out the Technical Report.
License/Terms of Use:
GOVERNING TERMS: Use of this model is governed by the CC-BY-4.0 license.
Discover more from NVIDIA:
For documentation, deployment guides, enterprise-ready APIs, and the latest open models—including Nemotron and other cutting-edge speech, translation, and generative AI—visit the NVIDIA Developer Portal at developer.nvidia.com.
Join the community to access tools, support, and resources to accelerate your development with NVIDIA’s NeMo, Riva, NIM, and foundation models.
Figure 1: ASR WER comparison across different models. This does not include Punctuation and Capitalisation errors.
Evaluation Notes
Note 1: The above evaluations are conducted for 24 supported languages, excluding Latvian since seamless-m4t-v2-large and seamless-m4t-medium do not support it.
Note 2: Performance differences may be partly attributed to Portuguese variant differences - our training data uses European Portuguese while most benchmarks use Brazilian Portuguese.
Deployment Geography:
Global
Use Case:
This model serves developers, researchers, academics, and industries building applications that require speech-to-text capabilities, including but not limited to: conversational AI, voice assistants, transcription services, subtitle generation, and voice analytics platforms.
This model was developed based on FastConformer encoder architecture[3] and TDT decoder[4]
This model has 600 million model parameters.
Input:
Input Type(s): 16kHz Audio
Input Format(s):.wav and .flac audio formats
Input Parameters: 1D (audio signal)
Other Properties Related to Input: Monochannel audio
Output:
Output Type(s): Text
Output Format: String
Output Parameters: 1D (text)
Other Properties Related to Output: Punctuations and Capitalizations included.
Our AI models are designed and/or optimized to run on NVIDIA GPU-accelerated systems. By leveraging NVIDIA's hardware (e.g. GPU cores) and software frameworks (e.g., CUDA libraries), the model achieves faster training and inference times compared to CPU-only solutions.
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 -U nemo_toolkit['asr']
The model is available for use in the NeMo toolkit [5], and can be used as a pre-trained checkpoint for inference or for fine-tuning on another dataset.
Automatically instantiate the model
python
1import nemo.collections.asr as nemo_asr
2asr_model = nemo_asr.models.ASRModel.from_pretrained(model_name="nvidia/parakeet-tdt-0.6b-v3")
1output = asr_model.transcribe(['2086-149220-0033.wav'], timestamps=True)2# by default, timestamps are enabled for char, word and segment level3word_timestamps = output[0].timestamp['word']# word level timestamps for first sample4segment_timestamps = output[0].timestamp['segment']# segment level timestamps5char_timestamps = output[0].timestamp['char']# char level timestamps67for stamp in segment_timestamps:8print(f"{stamp['start']}s - {stamp['end']}s : {stamp['segment']}")
Transcribing long-form audio
python
1#updating self-attention model of fast-conformer encoder2#setting attention left and right context sizes to 2563asr_model.change_attention_model(self_attention_model="rel_pos_local_attn", att_context_size=[256,256])45output = asr_model.transcribe(['2086-149220-0033.wav'])67print(output[0].text)
Streaming with Parakeet models
To use parakeet models in streaming mode use this script as shown below:
bash
1python NeMo/main/examples/asr/asr_chunked_inference/rnnt/speech_to_text_streaming_infer_rnnt.py \2pretrained_name="nvidia/parakeet-tdt-0.6b-v3"\3model_path=null \4audio_dir="<optional path to folder of audio files>"\5dataset_manifest="<optional path to manifest>"\6output_filename="<optional output filename>"\7right_context_secs=2.0\8chunk_secs=2\9left_context_secs=10.0\10batch_size=32\11clean_groundtruth_text=False
Atleast 2GB RAM for model to load. The bigger the RAM, the larger audio input it supports.
Model Version
Current version: parakeet-tdt-0.6b-v3. Previous versions can be accessed here.
Training and Evaluation Datasets:
Training
This model was trained using the NeMo toolkit [5], following the strategies below:
Initialized from a CTC multilingual checkpoint pretrained on the Granary dataset [1] [2].
Trained for 150,000 steps on 128 A100 GPUs.
Dataset corpora and languages were balanced using a temperature sampling value of 0.5.
Stage 2 fine-tuning was performed for 5,000 steps on 4 A100 GPUs using approximately 7,500 hours of high-quality, human-transcribed data of NeMo ASR Set 3.0.
During the training, a unified SentencePiece Tokenizer [6] with a vocabulary of 8,192 tokens was used. The unified tokenizer was constructed from the training set transcripts using this script and was optimized across all 25 supported languages.
All transcriptions preserve punctuation and capitalization. The Granary dataset will be made publicly available after presentation at Interspeech 2025.
Data Collection Method by dataset
Hybrid: Automated, Human
Labeling Method by dataset
Hybrid: Synthetic, Human
Properties:
Noise robust data from various sources
Single channel, 16kHz sampled data
Evaluation Datasets
For multilingual ASR performance evaluation:
Fleurs [10]
MLS [11]
CoVoST [12]
For English ASR performance evaluation:
Hugging Face Open ASR Leaderboard [13] datasets
Data Collection Method by dataset
Human
Labeling Method by dataset
Human
Properties:
All are commonly used for benchmarking English ASR systems.
Audio data is typically processed into a 16kHz mono channel format for ASR evaluation, consistent with benchmarks like the Open ASR Leaderboard.
Performance
Multilingual ASR
The tables below summarizes the WER (%) using a Transducer decoder with greedy decoding (without an external language model):
Language
Fleurs
MLS
CoVoST
Average WER ↓
11.97%
7.83%
11.98%
bg
12.64%
-
-
cs
11.01%
-
-
da
18.41%
-
-
de
5.04%
-
4.84%
el
20.70%
-
-
en
4.85%
-
6.80%
es
3.45%
4.39%
3.41%
et
17.73%
-
22.04%
fi
13.21%
-
-
fr
5.15%
4.97%
6.05%
hr
12.46%
-
-
hu
15.72%
-
-
it
3.00%
10.08%
3.69%
lt
20.35%
-
-
lv
22.84%
-
38.36%
mt
20.46%
-
-
nl
7.48%
12.78%
6.50%
pl
7.31%
7.28%
-
pt
4.76%
7.50%
3.96%
ro
12.44%
-
-
ru
5.51%
-
3.00%
sk
8.82%
-
-
sl
24.03%
-
31.80%
sv
15.08%
-
20.16%
uk
6.79%
-
5.10%
Note: WERs are calculated after removing Punctuation and Capitalization from reference and predicted text.
NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications. When downloaded or used in accordance with our terms of service, developers should work with their supporting model team to ensure this model meets requirements for the relevant industry and use case and addresses unforeseen product misuse.
For more detailed information on ethical considerations for this model, please see the Model Card++ Explainability, Bias, Safety & Security, and Privacy Subcards here.
Please report security vulnerabilities or NVIDIA AI Concerns here.
Bias:
Field
Response
Participation considerations from adversely impacted groups protected classes in model design and testing
None
Measures taken to mitigate against unwanted bias
None
Explainability:
Field
Response
Intended Domain
Speech to Text Transcription
Model Type
FastConformer
Intended Users
This model is intended for developers, researchers, academics, and industries building conversational based applications.
Output
Text
Describe how the model works
Speech input is encoded into embeddings and passed into conformer-based model and output a text response.
Name the adversely impacted groups this has been tested to deliver comparable outcomes regardless of
Not Applicable
Technical Limitations & Mitigation
Transcripts may be not 100% accurate. Accuracy varies based on language and characteristics of input audio (Domain, Use Case, Accent, Noise, Speech Type, Context of speech, etc.)
Verified to have met prescribed NVIDIA quality standards
Yes
Performance Metrics
Word Error Rate
Potential Known Risks
If a word is not trained in the language model and not presented in vocabulary, the word is not likely to be recognized. Not recommended for word-for-word/incomplete sentences as accuracy varies based on the context of input text
Licensing
GOVERNING TERMS: Use of this model is governed by the CC-BY-4.0 license.
Privacy:
Field
Response
Generatable or reverse engineerable personal data?
None
Personal data used to create this model?
None
Is there provenance for all datasets used in training?
Yes
Does data labeling (annotation, metadata) comply with privacy laws?
Yes
Is data compliant with data subject requests for data correction or removal, if such a request was made?
The Principle of least privilege (PoLP) is applied limiting access for dataset generation and model development. Restrictions enforce dataset access during training, and dataset license constraints adhered to.
