ContainersNVIDIADeepSeek-R1NVIDIALlama-3.1-Nemotron-70B-InstructNVIDIAPyTorch
CollectionsNVIDIAOmniverse Kit (FB)NVIDIADeepStream SDKNVIDIAOmniverse Kit App Streaming
ModelsNVIDIAStyleGAN3 pretrained modelsNVIDIAPeopleNetNVIDIATrafficCamNet
ResourcesNVIDIARiva Skills Quick Start
Helm ChartsNVIDIAGPU OperatorNVIDIANVIDIA NIM Operator
  • NVIDIA
    RIVA Parakeet-RNNT-XXL-1.1B ASR Multilingual Universal Tokenizer
    Model
    NVIDIA
    RIVA Parakeet-RNNT-XXL-1.1B ASR Multilingual Universal Tokenizer

    Multilingual Parakeet-RNNT-XXL-1.1B ASR model with Universal Tokenizer trained on ASR set 1.0

    This model is backed by NVIDIA's Plus Plus (++) Promise
    to learn more about the quality of the datasets used to train this model.
    FieldResponse
    What is the language balance of the model validation data?en-US: 35.0%, es-US: 9.66%, ko-KR: 8.07%, ru-RU: 6.33%, ar-AR: 5.65%, de-DE: 5.2%, fr-FR: 4.6%, hi-IN: 4.59%, it-IT: 4.2%, pt-BR: 3.5%, ja-JP: 2.82%, es-ES: 1.9%, pt-PT: 1.5%, nl-NL: 1.4%, tr-TR: 1.3%, fr-CA: 1.1%, en-GB: 0.95%, pl-PL: 0.64%, th-TH: 0.59%, he-IL: 0.26%, cs-CZ: 0.21%, da-DK: 0.19%, nb-NO: 0.18%, sv-SE: 0.16%
    What is the geographic origin language balance of the model validation data?North America: 45.75%, Europe: 27.47%, Asia: 16.1%, Middle East: 7.2%, South America: 3.5%
    What is the accent balance of the model validation data?en-US: 35.0%, es-US: 9.66%, ko-KR: 8.07%, ru-RU: 6.33%, ar-AR: 5.65%, de-DE: 5.2%, fr-FR: 4.6%, hi-IN: 4.59%, it-IT: 4.2%, pt-BR: 3.5%, ja-JP: 2.82%, es-ES: 1.9%, pt-PT: 1.5%, nl-NL: 1.4%, tr-TR: 1.3%, fr-CA: 1.1%, en-GB: 0.95%, pl-PL: 0.64%, th-TH: 0.59%, he-IL: 0.26%, cs-CZ: 0.21%, da-DK: 0.19%, nb-NO: 0.18%, sv-SE: 0.16%
    Participation considerations from adversely impacted groups (protected classes) in model design and testing:Age, Gender, Linguistic Background
    Measures taken to mitigate against unwanted bias:Used custom dataset to validate model performance across gender, age, and linguistic demographics