Skip to content
Language models

univ_full_eres2conformer_xl

flagship model for speaker verification and recognition

2 min read

The new univ_full_eres2conformer_xl model from the FlaxSpeaker library solves the task of voice identification and verification from audio recordings. It is built on the ERes2Conformer (XL) architecture with catp pooling (Attentive Temporal Pooling — a weighted averaging mechanism for temporal features in speech) and trained on a massive universal dataset of 12,193 speakers and 1.7 million clean utterances. The model outputs a compact 256-dimensional feature vector (embedding), which can be used to accurately determine whether a voice belongs to a specific person. The model is useful for developers of voice systems, biometric services, and audio analysis applications.

Key features

  • High accuracy: low Equal Error Rate (EER) on the VoxCeleb1 and LibriSpeech benchmarks.
  • Audio frontend: uses 80-dimensional mel-spectrograms with Voice Activity Detection (VAD) and Cepstral Mean and Variance Normalization (CMVN).
  • Multi-corpus training: the training set includes VoxCeleb, CN-Celeb, and LibriSpeech with noise and speed perturbation augmentations.
  • Efficiency: contains about 64.3 million parameters in inference mode and supports export to TFLite.

What you can do

  • Speaker verification: compare two audio recordings to check if the same person is speaking.
  • Embedding extraction: get a vector representation of a voice from a FLAC or WAV file for clustering or search.
  • Biometric protection: apply the model in voice-based access control systems for mobile or server applications.
  • Pipeline integration: use the flaxspeaker library in Python scripts for streaming or file audio processing.

Example

Code example for loading the model and verifying a voice via the Python API:

from flaxspeaker import load_pretrained

model = load_pretrained("wq2012/univ_full_eres2conformer_xl")
emb = model.extract_embedding("path/to/utterance.flac")
score = model.verify("path/to/enroll.flac", "path/to/test.flac")
print(f"Embedding shape: {emb.shape}, Verification score: {score:.4f}")
Read moreDetails
Benchmarks: VoxCeleb1-O EER 0.44%, LibriSpeech test-clean EER 0.52% (single-utterance mode).
Parameters: 64,376,300 inference parameters (encoder + scoring heads).
Training: 145,000 gradient descent steps with a batch size of 512, seed 42.
Requirements: flaxspeaker >= 0.7.0 library, JAX/Flax support.
Discuss Telegram 0

Will it run?

Runs in F32

Fits without quantisation.

  • F32weights 0.2 GB · ready files0.9 GB
  • BF16weights 0.1 GB0.7 GB
  • FP8 / INT8recommendedweights 0.1 GB0.7 GB
  • 4-bitweights 0.0 GB0.6 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · F32 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

A ready script for the machine we advise above: paste it and the model downloads and starts.

  1. 1. Rent A30

    $0.26 an hour with the coupon. Pick the Ubuntu image when you create the machine.

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # wq2012/univ_full_eres2conformer_xl — A30, FP8 / INT8
    # NVIDIA driver: CUDA 13 needs 580+. If older or missing, the current nvidia-open is installed (keep yours: KEEP_DRIVER=1)
    DRV=$(nvidia-smi --query-gpu=driver_version --format=csv,noheader 2>/dev/null | head -1 | cut -d. -f1)
    case "$DRV" in ""|*[!0-9]*) DRV=0 ;; esac
    if [ "$DRV" -lt 580 ] && [ -z "${KEEP_DRIVER:-}" ]; then
      . /etc/os-release && \
      curl -fsSLO "https://developer.download.nvidia.com/compute/cuda/repos/ubuntu${VERSION_ID//./}/x86_64/cuda-keyring_1.1-1_all.deb" && \
      sudo dpkg -i cuda-keyring_1.1-1_all.deb && rm -f cuda-keyring_1.1-1_all.deb && \
      sudo apt-get update && \
      sudo DEBIAN_FRONTEND=noninteractive apt-get install -y nvidia-open || \
      { echo "Could not install the NVIDIA driver: see the error above. To keep the current one, run with KEEP_DRIVER=1."; exit 1; }
      echo "NVIDIA driver updated. Reboot the machine (sudo reboot) and run this same script again."
      exit 0
    fi
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv venv ~/vllm --python 3.12 && source ~/vllm/bin/activate
    uv pip install vllm
    vllm serve wq2012/univ_full_eres2conformer_xl --max-model-len 8192 --quantization fp8
  3. 3. Use it

    The server listens on port 8000 and speaks the OpenAI API: http://localhost:8000/v1 on the machine. From your computer: ssh -L 8000:localhost:8000 user@machine-IP.

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

  • MTP (multi-token prediction)0.34.0from November 16, 2026new release

    1.5–2× generation speed

    If the model has MTP layers (DeepSeek-V3/R1, Qwen3-Next, GLM, MiMo and others), it predicts several tokens per step itself.

    How to enable: vLLM / SGLang: --speculative-config '{"method": "mtp", "num_speculative_tokens": 1}'. Check the model card.

    Earlier generations (2)
    • EAGLE-3 — Speculative decoding with a small draft head. For models without their own MTP layers.
    • EAGLE-2 / Medusa — Earlier draft-head generations. Still work if that is all you find.
  • FP8 KV cache0.34.0from November 16, 2026new release

    2× more context in the same memory

    Keeps the attention cache in 8 bits, so long contexts fit where they did not.

    How to enable: vLLM: --kv-cache-dtype fp8; llama.cpp: -ctk q8_0 -ctv q8_0.

  • vLLM / SGLang0.34.0from November 16, 2026new release

    many times more tokens/s on a server

    Server engines: batching, FlashAttention, FP8 and speculative decoding out of the box.

    How to enable: vllm serve <model> — an OpenAI-compatible API.

    Earlier generations (1)
    • llama.cpp — GGUF models on any GPU, Mac or even CPU; some layers can stay in RAM.

Model family

New posts about versions of univ_full_eres2conformer_xl (quants, finetunes, LoRAs) will show up under the bell.

No known relatives of this model yet. All derivatives on Hugging Face

About the model

wq2012/univ_full_eres2conformer_xl
Author
wq2012
Category
Language models
Released
Posted

On NeyroLearn

Hugging Face

00

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments