GGUF quantizations of the CED family (Consistent Ensemble Distillation, Xiaomi) - SOTA-tier audio-tagging models that classify everyday sounds (baby cry, footsteps, glass breaking, alarms, dog bark,...) into the 527-class AudioSet ontology. These files run with ced.cpp, a standalone C++/ggml port (no Python, no PyTorch at inference), and with LocalAI via the ced backend. Converted from the mispeech/ced- checkpoints (Apache-2.0). CED is a plain AST/DeiT Vision Transformer over a log-mel spectrogram; the port is numerically equal to the PyTorch reference. One self-contained GGUF per size + quant (config, 527 labels, and the mel filterbank/window are all embedded). Pick by your accuracy/size…
Open-weight model · Audio classification
Silero-VAD-v6.2.1-MLX
by Ivan aufklarer/Silero-VAD-v6.2.1-MLX
MLX port of snakers4/silero-vad tag v6.2.1 for voice activity detection on Apple Silicon. Measured with speech-swift release tests on Apple Silicon using a 20 s 16 kHz speech fixture, 625 streaming chunks.
Parameters309,121
Context—
Weights1.2 MB
Licensemit
AccessOpen weights
Monthly Downloads15.4k
Runs On
What it takes to serve Silero-VAD-v6.2.1-MLX (309,121 parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 0.0 GB | 0.0 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 0.0 GB | 0.0 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 0.0 GB | 0.0 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Sep 18, 2026.
Model Card
By Ivan, published under mit, revision 0046cea48b26.
MLX port of snakers4/silero-vad tag v6.2.1 for voice activity detection on Apple Silicon. Measured with speech-swift release tests on Apple Silicon using a 20 s 16 kHz speech fixture, 625 streaming chunks. Parity against the matching CoreML v6.2.1 export: The exported safetensors were also checked tensor-by-tensor against the upstream v6.2.1 JIT state dict after conversion; the maximum absolute tensor difference was 0. Converted from snakers4/silero-vad tag v6.2.1. The upstream project is MIT licensed. - speech-swift - Apple SDK - Docs - install and CLI docs - soniqo.audio - website - blog - blog
Read Ivan's full model card
MLX port of snakers4/silero-vad
tag v6.2.1 for voice activity detection on Apple Silicon.
Model
| Field | Value |
|---|---|
| Parameters | 309,121 |
| Quantization | none, Float32 weights |
| Format | MLX safetensors |
| Sample rate | 16 kHz |
| Chunk size | 512 samples, 32 ms |
| Context | 64 samples |
| Upstream | snakers4/silero-vad:v6.2.1 |
Files
| File | Size | Description |
|---|---|---|
model.safetensors |
1.2 MB | MLX weights |
config.json |
456 B | Model metadata and runtime shape config |
README.md |
- | Model card |
Performance
Measured with speech-swift release tests on Apple Silicon using a 20 s
16 kHz speech fixture, 625 streaming chunks.
| Backend | Segment | Latency per chunk | RTF |
|---|---|---|---|
| Silero v6.2.1 MLX | 5.184s-8.416s | 0.4999 ms | 0.01562 |
| Silero v6.2.1 CoreML | 5.184s-8.416s | 0.0630 ms | 0.00197 |
Parity against the matching CoreML v6.2.1 export:
| Metric | Value |
|---|---|
| Max probability diff | 0.060569 |
| Average probability diff | 0.001673 |
The exported safetensors were also checked tensor-by-tensor against the upstream v6.2.1 JIT state dict after conversion; the maximum absolute tensor difference was 0.
Usage
Swift SDK
import SpeechVAD
let vad = try await SileroVADModel.fromPretrained(engine: .mlx)
let segments = vad.detectSpeech(audio: samples, sampleRate: 16000)
CLI
speech vad input.wav
Source
Converted from snakers4/silero-vad
tag v6.2.1. The upstream project is MIT licensed.
Links
- speech-swift - Apple SDK
- Docs - install and CLI docs
- soniqo.audio - website
- blog - blog
Configuration
- Model type
- silero_vad_mlx
Identity and Version
- Repository
- aufklarer/Silero-VAD-v6.2.1-MLX
- Publisher
- Ivan
- Task
- Audio classification
- Modality
- Audio
- Library
- mlx
- Parameters
- 309,121 parameters
- Languages
- mlx, vad
- Revision
- 0046cea48b26401909b24292b688fbc9b6322bc5
- First published
- 2026-06-28
- Last updated
- 2026-06-28
Files and Weights
4 files, 1.2 MB in total. The weights are 1 file totalling 1.2 MB in safetensors.
Weights1 file · 1.2 MB
Configuration1 file · 456 B
Documentation1 file · 2.1 KB
Repository1 file · 1.5 KB
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model.safetensors | Weights | 1.2 MB | 8367dac03e6c |
| config.json | Configuration | 456 B | — |
| README.md | Documentation | 2.1 KB | — |
| .gitattributes | Repository | 1.5 KB | — |
License and Download
- License
- mit
- Access
- Open weights, no gate
- Download size
- 1.2 MB
Download from Ivan
Released by Ivan through its official repository on Hugging Face. Read the license.
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 1.2 MB |
| 16-bit | 0.0 GB |
| 8-bit | 0.0 GB |
| 4-bit | 0.0 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About Silero-VAD-v6.2.1-MLX
How much GPU memory does Silero-VAD-v6.2.1-MLX need?
About 0 GB at 16-bit and 0 GB at 4-bit: the weights (309,121 parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run Silero-VAD-v6.2.1-MLX on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Can I use Silero-VAD-v6.2.1-MLX commercially?
Yes. Silero-VAD-v6.2.1-MLX is released under MIT License. The MIT License is a short permissive license. It permits commercial use, modification and redistribution, provided the copyright notice and permission notice are included.
Similar Models
This is a spoken language recognition model trained on the VoxLingua107 dataset using SpeechBrain. The model uses the ECAPA-TDNN architecture that has previously been used for speaker recognition. However, it uses more fully connected hidden layers after the embedding layer, and cross-entropy loss was used for training. We observed that this improved the performance of extracted utterance embeddings for downstream tasks. The system is trained with recordings sampled at 16kHz (single channel). The code will automatically normalize your audio (i.e., resampling + mono channel selection) when calling classifyfile if needed. The model can classify a speech utterance according to the language…
Abstract: The recognition of accented speech still remains a dominant problem in Automatic Speech Recognition (ASR) systems. We approach the classification of accented English speech through the Emphasized Channel Attention, Propagation and Aggregation Time Delay Neural Network (ECAPA-TDNN) architecture which has been shown to perform well on a variety of speech tasks. Three models are proposed: one trained from scratch, another two models (one using data augmentation and a baseline model) fine-tuned from the checkpoints of speechbrain/spkrec-ecapa-voxceleb (VoxCeleb). Our results show that the model fine-tuned with data augmentation yield the best results. Most of the misclassifications…
The development log of our Music Audio Pre-training (m-a-p) model family: - 02/06/2023: arxiv pre-print and training codes released. - 17/03/2023: we release two advanced music understanding models, MERT-v1-95M and MERT-v1-330M, trained with new paradigm and dataset. They outperform the previous models and can better generalize to more tasks. - 14/03/2023: we retrained the MERT-v0 model with open-source-only music dataset MERT-v0-public - 29/12/2022: a music understanding model MERT-v0 trained with MLM paradigm, which performs better at downstream tasks. - 29/10/2022: a pre-trained MIR model music2vec trained with BYOL paradigm. Here is a table for quick model pick-up: The m-a-p models…
The development log of our Music Audio Pre-training (m-a-p) model family: - 02/06/2023: arxiv pre-print and training codes released. - 17/03/2023: we release two advanced music understanding models, MERT-v1-95M and MERT-v1-330M, trained with new paradigm and dataset. They outperform the previous models and can better generalize to more tasks. - 14/03/2023: we retrained the MERT-v0 model with open-source-only music dataset MERT-v0-public - 29/12/2022: a music understanding model MERT-v0 trained with MLM paradigm, which performs better at downstream tasks. - 29/10/2022: a pre-trained MIR model music2vec trained with BYOL paradigm. Here is a table for quick model pick-up: The m-a-p models…
This model combines the SpeechBrain ECAPA-TDNN speaker embedding model with an SVM classifier to predict speaker gender from audio input. The model was trained and evaluated on the VoxCeleb2, Mozilla Common Voice v10.0, and TIMIT datasets - Mozilla Common Voice v10.0 English validated test set: 92.3% accuracy The model was trained on VoxCeleb2 dataset: - Converted to WAV format, single channel, 16kHz sampling rate, 256 kp/s bitrate - Applied SileroVAD for voice activity detection, taking the first voiced segment You can install the package directly from GitHub: - Model was trained on celebrity voices from YouTube interviews - Performance may vary on different audio qualities or recording…