--- license: other license_name: nvidia-software-and-model-evaluation-license license_link: >- https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-software-and-model-evaluation-license/ --- # Nemotron-3.5-asr-streaming-0.6b [![Model architecture](https://img.shields.io/badge/Model_Arch-FastConformer--CacheAware--RNNT-lightgrey#model-badge)](#model-architecture) | [![Model size](https://img.shields.io/badge/Params-600M-lightgrey#model-badge)](#model-architecture) > [!Note] > May 29, 2026: nemotron-3.5-asr-streaming-0.6b checkpoint was udpated. # Model Overview ### Description: **Nemotron 3 ASR Multilingual (Nemotron-ASR-Streaming-Multilingual)** is a multilingual, streaming Automatic Speech Recognition (ASR) engineered to deliver high-quality multilingual transcription across both low-latency streaming and high-throughput batch workloads. Developed by NVIDIA, this 600M parameter model transcribes speech into text with native support for punctuation and capitalization.
By leveraging a state-of-the-art **Cache-Aware FastConformer-RNNT** architecture, the model eliminates redundant overlapping computations common in traditional "buffered" streaming. This allows it to process only new audio chunks while reusing cached encoder context, significantly improving computational efficiency and minimizing end-to-end delay without sacrificing accuracy.
The model transcribes the following languages: English (en-US, en-GB), Spanish (es-US, es-ES), German (de-DE), French (fr-FR, fr-CA), Italian (it-IT), Arabic (ar-AR), Japanese (ja-JP), Korean (ko-KR), Portuguese (pt-BR, pt-PT), Russian (ru-RU), Hindi (hi-IN), Mandarin (zh-CN), Vietnamese (vi-VN), Hebrew (he-IL), Dutch (nl-NL), Czech (cz-CZ), Danish (da-DK), Polish (pl-PL), Norwegian (nn-NO, nb-NO), Swedish (sv-SE), Thai (th-TH), Turkish (tr-TR), Bulgarian (bg-BG), Greek (el-GR), Estonian (et-EE), Finnish (fi-FI), Croatian (hr-HR), Hungarian (hu-HU), Lithuanian (lt-LT), Latvian (lv-LV), Romanian (ro-RO), Slovak (sk-SK), Ukrainian (uk-UA), Maltese (mt-MT), and Slovenian (sl-SL). It supports uppercase and lowercase letters, punctuation, spaces, and apostrophes. It was trained on a massive ASR dataset of approximately 450,000 hours of speech and is engineered to perform across diverse and challenging acoustic conditions. **This model is ready for commercial/non-commercial use.**
### License/Terms of Use: Governing Terms: Use of this model is governed by the [NVIDIA Model Evaluation License Agreement](https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-software-and-model-evaluation-license/). ### Deployment Geography: Global
### Use Case:
This model is for transcription of multilingual audio.
### Release Date:
Hugging Face [06/05/2026] via [https://huggingface.co/nvidia/nemotron-asr-streaming-multilingual-0.6b]
## References(s): [1] [Stateful Conformer with Cache-based Inference for Streaming Automatic Speech Recognition](https://arxiv.org/pdf/2312.17279)
[2] [Fast Conformer with Linearly Scalable Attention for Efficient Speech Recognition](https://arxiv.org/abs/2305.05084) ## Model Architecture: **Architecture Type:** FastConformer-CacheAware-RNNT with Prompt
The model is based on the Cache-Aware [1] FastConformer [2] architecture with 24 encoder layers and an RNNT (Recurrent Neural Network Transducer) decoder. The cache-aware streaming design enables efficient processing of audio in chunks while maintaining context from previous frames. Unlike buffered inference, this model maintains caches for all encoder self-attention and convolution layers. This enables reuse of hidden states at every streaming step, where cached activations eliminate redundant computations. As a result, there are no overlapping computations; each processed frame is strictly non-overlapping. This model leverages prompts to guide the transcription process, enabling language-specific transcription from a single ASR model through language ID conditioning.
**Network Architecture:** * Encoder: Cache-Aware FastConformer with 24 layers * Decoder: RNNT (Recurrent Neural Network Transducer) **This model was developed based on [nvidia/nemotron-speech-streaming-en-0.6b](https://huggingface.co/nvidia/nemotron-speech-streaming-en-0.6b) [1].
** Number of model parameters: 600M
## Input(s):
**Input Type(s):** Audio, Lang ID
**Input Format(s):** wav, string
**Input Parameters:** One-Dimensional (1D) for audio and One-Dimensional (1D) for Lang ID
**Other Properties Related to Input:** Maximum Length in seconds specific to GPU Memory, No Pre-Processing Needed, Mono channel is required. 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.
## Output(s) **Output Type(s):** Text String in Input Language
**Output Format(s):** String
**Output Parameters:** One-Dimensional (1D)
**Other Properties Related to Output:** No Maximum Character Length, transcribe punctuation and capitalization. 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.
## How to use this model **NeMo-framework (EA):**
```bash git clone https://github.com/kingformatty/NeMo.git git checkout prompt_unitifed_architecture_hf_EA ``` **Streaming inference with target language:**
Supported att_context_size: [56,0] (default), [56,3], [56,6], [56,13] ```bash python ${NEMO_ROOT}/examples/asr/asr_cache_aware_streaming/speech_to_text_cache_aware_streaming_infer.py \ model_path=${MODEL_PATH} \ dataset_manifest=${MANIFEST_PATH} \ output_path=${OUTPUT_FOLDER} \ target_lang=es-ES \ att_context_size="[56,0]" \ strip_lang_tags=true ``` (es-ES for example, supported languages can be found in model description) **Streaming inference with auto model (agnostic language):**
```bash python ${NEMO_ROOT}/examples/asr/asr_cache_aware_streaming/speech_to_text_cache_aware_streaming_infer.py \ model_path=${MODEL_PATH} \ dataset_manifest=${MANIFEST_PATH} \ output_path=${OUTPUT_FOLDER} \ target_lang=auto \ att_context_size="[56,0]" \ strip_lang_tags=true ``` ## Software Integration: **Runtime Engine(s):** * NeMo 25.11, Riva 2.25.0 or higher
**Supported Hardware Microarchitecture Compatibility:**
* NVIDIA Ampere
* NVIDIA Blackwell
* NVIDIA Jetson
* NVIDIA Hopper
* NVIDIA Lovelace
* NVIDIA Turing
* NVIDIA Volta
**Supported Operating System(s):** * Linux
* Linux 4 Tegra
The integration of foundation and fine-tuned models into AI systems requires additional testing using use-case-specific data to ensure safe and effective deployment. Following the V-model methodology, iterative testing and validation at both unit and system levels are essential to mitigate risks, meet technical and functional requirements, and ensure compliance with safety and ethical standards before deployment. ## Model Version(s): nemotron-asr-streaming-multilingual-v1
## Training and Evaluation Datasets: ## Training Dataset ** Data Modality: Audio
** Audio Training Data Size: 10,000 to 1 Million Hours
** Data Collection Method by dataset
* Human
** Labeling Method by dataset
* Human
* Synthetic: labels generated from Parakeet-CTC-XXL-1.1b model, PnC generated from [Qwen3-32B](https://huggingface.co/Qwen/Qwen3-32B)
**Properties:** In excess of 450,000 hours of speech across the following languages: English (en-US, en-GB), Spanish (es-US, es-ES), German (de-DE), French (fr-FR, fr-CA), Italian (it-IT), Arabic (ar-AR), Japanese (ja-JP), Korean (ko-KR), Portuguese (pt-BR, pt-PT), Russian (ru-RU), Hindi (hi-IN), Mandarin (zh-CN), Vietnamese (vi-VN), Hebrew (he-IL), Dutch (nl-NL), Czech (cz-CZ), Danish (da-DK), Polish (pl-PL), Norwegian (nn-NO, nb-NO), Swedish (sv-SE), Thai (th-TH), Turkish (tr-TR), Bulgarian (bg-BG), Greek (el-GR), Estonian (et-EE), Finnish (fi-FI), Croatian (hr-HR), Hungarian (hu-HU), Lithuanian (lt-LT), Latvian (lv-LV), Romanian (ro-RO), Slovak (sk-SK), Ukrainian (uk-UA), Maltese (mt-MT), and Slovenian (sl-SL). The training data is a dynamic blend of public and proprietary internal datasets normalized to have spoken forms in text with punctuation and capitalization.
## Evaluation Dataset ** Data Collection Method by dataset
* Human
** Labeling Method by dataset
* Human
**Properties:** A dynamic blend of public and internal proprietary normalized to spoken forms in text with punctuation and capitalization.
# Inference: **Acceleration Engine:** Triton
**Test Hardware:**
* NVIDIA Blackwell
* NVIDIA A10
* NVIDIA A100
* NVIDIA A30
* NVIDIA H100
* NVIDIA L4
* NVIDIA L40
## Ethical Considerations: 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 internal 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++ Bias, Explainability, Safety & Security, and Privacy Subcards](Insert Link Here).
Please report model quality, risk, security vulnerabilities or NVIDIA AI Concerns [here](https://www.nvidia.com/en-us/support/submit-security-vulnerability/).
## Get Help ### Enterprise Support Get access to knowledge base articles and support cases or [submit a ticket](https://www.nvidia.com/en-us/data-center/products/ai-enterprise-suite/support/).