⌘KCtrl+K
Enhance input speech recorded with low-quality microphones in noisy or reverberant environments, producing studio-quality speech.
Studio Voice is a speech enhancement model from NVIDIA. Studio Voice enhances input speech recorded with low-quality microphones in noisy or reverberant environments, producing studio-quality speech.
This model is ready for commercial use.
Use of this model is governed by the NVIDIA Open Model License.
Global
Studio Voice models removes microphone artifacts and room reverberations to produce studio quality voice. It is intended to be used by content developers and broadcasters.
NGC [03/16/2026] via afx_studio_voice
Architecture Type: Convolution Neural Networks (CNNs), Transformers, Generative Adversarial Networks (GANs)
Network Architecture: Encoder-Decoder
Number of model parameters: 183M
Input Type(s): Audio
Input Format(s): PCM F32
Input Parameters: One-Dimensional (1D)
Other Properties Related to Input: Pulse Code Modulation (PCM) audio samples
with no encoding or pre-processing; 16kHz or 48kHz sampling rate required.
Output Type(s): Audio
Output Format: PCM F32
Output Parameters: One-Dimensional (1D)
Other Properties Related to Output: PCM audio samples at input sampling rate
with no encoding or post-processing.
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.
Runtime Engine(s):
Supported Hardware Microarchitecture Compatibility:
Preferred/Supported Operating System(s):
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.
This AI model can be embedded as an Application Programming Interface (API) call into the software environment described above.
StudioVoice v0.5.1
Data Modality:
Audio Training Data Size:
Dataset partition: Training [80%], Testing [10%], Validation [10%]
NVIDIA models are trained on a diverse set of public and proprietary datasets. The Studio Voice model is trained on a dataset that comprises of diverse English accents and different types of microphone devices.
Data Collection Method by dataset: [Hybrid: Human, Synthetic] Labeling Method by dataset: [Hybrid: Human, Synthetic]
Link: DAPS
Properties:
The DAPS dataset has 15 versions of audio (3 professional versions and
12 consumer device/real-world environment combinations). Each version consists
of about 4.5 hours of data (about 14 minutes from each of 20 speakers).
Link: LibriTTS
Properties:
LibriTTS is a multi-speaker English corpus of approximately 585 hours of read
English speech, which is resampled at 16kHZ.
Link: VCTK
Properties:
This CSTR VCTK Corpus includes speech data uttered by 110 English speakers with
various accents. Each speaker reads out about 400 sentences, which were selected
from a newspaper, the rainbow passage and an elicitation paragraph used for the
speech accent archive.
Link: HiFi-TTS
Properties:
A multi-speaker English dataset for training text-to-speech models.
The HiFi-TTS dataset contains about 291.6 hours of speech from 10 speakers with
at least 17 hours per speaker sampled at 44.1 kHz.
Link: Device Recorded VCTK (DR-VCTK)
Properties:
Device recorded version of VCTK dataset on common consumer devices
(laptop, tablet and smartphone) in office environment. This dataset contains
109 English speakers with different accents. There are around 400 sentences
available from each speaker. For this recording, 8 different microphones were
used. This dataset contains around 250 Gb of re-recorded speech.
Link: Dataset of impulse responses from variable acoustics room Arni at Aalto Acoustic Labs
Properties:
A dataset of impulse responses collected in the variable acoustics laboratory
Arni at Acoustics Lab of Aalto University, Espoo, Finland. IRs of 5342
configurations of sound absorption in Arni are included in the dataset. Each of
them were measured using an omnidirectional sound source and 5 sound receivers.
For each configuration, 5 impulse reponses (IRs) were captured. The total number
of measurements in the dataset is 132 037.
Link: Room Impulse Response and Noise Database
Properties:
A database of simulated and real room impulse responses, isotropic and
point-source noises. The audio files in this data are all in 16KHz sampling rate
and 16-bit precision.
Link: DNS Challenge 5
Properties:
Collated dataset of clean speech, noise and impulse response provided by
Microsoft for the ICASSP 2023 Deep Noise Suppression Challenge.
Link: AudioSet
Properties:
AudioSet consists of an expanding ontology of 632 audio event classes and
a collection of 2,084,320 human-labeled 10-second sound clips drawn from
publicly available internet scale data.
Link: Multi-Language and Emotions Speech Dataset
Properties:
Multi-Language and Emotions Speech Dataset contains high quality
speech data worth approximately 140 hours and contains approximately 80 unique
speakers. The dataset contains different english accents and contains
different emotions as well. This dataset is taken from publicly available
internet scale data.
Link: Audio2Gesture Bones Dataset
Properties:
Audio2Gesture Bones Dataset is a speech dataset which is bought from Bones.studio
company. It contains only 2 speakers covering all different types of emotions.
It is a small dataset of around 4 GB and contains 7 hours of speech data.
Data Collection Method by dataset: [Hybrid: Human, Synthetic]
Labeling Method by dataset: [Hybrid: Human, Synthetic]
Properties:
The Studio Voice model is tested on a dataset that comprises of diverse English accents and different types of microphone devices. Test data is taken by sampling 10% of the training dataset mentioned above. The modality and data type is same as that of the training dataset.
Data Collection Method by dataset: [Hybrid: Human, Synthetic]
Labeling Method by dataset: [Hybrid: Human, Synthetic]
Properties:
The Studio Voice model is evaluated on a dataset that comprises of diverse English accents and different types of microphone devices. Evaluation data is taken by sampling 10% of the training dataset mentioned above. The modality and data type is same as that of the training dataset.
Acceleration Engine: Tensor(RT), Triton
Test Hardware:
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.
Please report model quality, risk, security vulnerabilities or NVIDIA AI Concerns here.