NGC Catalog
CLASSIC
Welcome Guest
Models
Riva ASR Mandarin-English LM

Riva ASR Mandarin-English LM

For downloads and more information, please view on a desktop device.
Logo for Riva ASR Mandarin-English LM
Description
Base Multilingual Code Switch Mandarin-English 4-gram LM
Publisher
NVIDIA
Latest Version
deployable_v3.0
Modified
March 19, 2025
Size
5.57 GB

Speech Recognition: Multilingual Code Switch Mandarin-English N-Gram Language Models

Description

Riva ASR Mandarin Unified LM estimates the likelihood of Mandarin (zh-CN)-English (en-US) code-switching (whether there is a mix of English and Mandarin in the word sequence), including capitalized words and punctuation symbols. When deployed, the ASR engine can optionally condition the transcript output on n-gram language models.

This model is ready for commercial use.

License/Terms of Use

GOVERNING TERMS: The use of this model is governed by the NVIDIA Community Model License (found at https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-community-models-license/).

References

KenLM

Model Architecture

Architecture Type: n-gram
Network Architecture: 4-gram trained with Kneser-Ney smoothing

Input

Input Type(s): Text in Mandarin and/or English
Input Format(s): String
Input Parameters: 1-Dimension Other Properties Related to Input: Sentences of any length; upper-cased, lower-cased, and punctuated text can be provided.

Output

Output Type(s): Likelihood of word sequence
Output Format: Float
Output Parameters: 1-Dimension
Other Properties Related to Output: The output represents the log-probability of any given sentence under the loaded language model.

How to Use this Model

The Riva Quick Start Guide is recommended as the starting point for trying out Riva models. For more information on using this model with Riva Speech Services, see the Riva User Guide.

There are a variety of formats contained within this model archive:

ARPA-formatted Language Models:

  • zh-CN_unified_3.0.arpa (For Punctuation & Capitalization Unified model)

KenLM-formatted Binary Language Models

  • zh-CN_unified_3.0.bin (For Punctuation & Capitalization Unified model)

Flashlight Decoder Vocabulary Files

  • zh-CN_unified_3.0_dict_vocab.txt (For Punctuation & Capitalization Unified model)

ARPA and KenLM binary formatted files can be used directly by the CTC CPU Decoder.

Software Integration

Runtime Engine(s):

  • Riva 2.19.0

Supported Hardware Microarchitecture Compatibility:

  • NVIDIA Ampere
  • NVIDIA Hopper
  • NVIDIA Jetson
  • NVIDIA Turing
  • NVIDIA Volta

[Preferred/Supported] Operating System(s):

  • Linux
  • Linux 4 Tegra

Model Version(s)

zh-CN_default_pnc_3.0

Training & Evaluation

Training Dataset

** Data Collection Method by dataset

  • Human

** Labeling Method by dataset

  • Human

Properties:

A dynamic blend of public and internal proprietary datasets normalized to have upper-cased, lower-cased, punctuated, and spoken forms in text.

Evaluation Dataset

** Data Collection Method by dataset

  • Human

** Labeling Method by dataset

  • Human

Properties:

A dynamic blend of public and internal proprietary datasets normalized to have upper-cased, lower-cased, punctuated, and spoken forms in text.

Inference

Engine: [Triton]
Test Hardware:

  • NVIDIA A10
  • NVIDIA A100
  • NVIDIA A30
  • NVIDIA H100
  • NVIDIA Jetson Orin
  • NVIDIA L4
  • NVIDIA L40
  • NVIDIA Turing T4
  • NVIDIA Volta V100

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 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.