This repository contains audio.cpp-native GGUF weights for Yue2-3B.
The audio.cpp code for Yue2 is ready on the
dev
branch
for community
testing, validation, and optimization. Star our repo so you don’t miss any important updates!
Files
File
Description
yue2-3b-bf16.gguf
Yue2-3B main model, BF16 weights.
yue2-3b-q8_0.gguf
Yue2-3B main model, Q8_0 quantized weights.
yue2-vae-f32.gguf
Yue2 VAE, F32 weights.
yue2-vae-f16.gguf
Yue2 VAE, F16 weights.
sidecars/yue2-model-config.json
Yue2-3B model config.
sidecars/yue2-generation-config.json
Yue2 generation defaults.
sidecars/yue2-qwen.tiktoken
Qwen tokenizer used by Yue2.
sidecars/yue2-vae-config.json
Yue2 VAE config.
examples/demo1.wav
Direct text-to-music demo,
cot=off
.
examples/demo1_q8.wav
Same demo generated with
yue2-3b-q8_0.gguf
and
yue2-vae-f16.gguf
.
examples/demo2.wav
Full planning demo,
cot=full
.
examples/demo2_q8.wav
Same demo generated with
yue2-3b-q8_0.gguf
and
yue2-vae-f16.gguf
.
examples/demo3.wav
Melody-conditioned cover demo,
cot=melody
.
examples/demo3_q8.wav
Same demo generated with
yue2-3b-q8_0.gguf
and
yue2-vae-f16.gguf
.
examples/demo4.wav
Full score-conditioned demo,
cot=full
.
examples/demo4_q8.wav
Same demo generated with
yue2-3b-q8_0.gguf
and
yue2-vae-f16.gguf
.
examples/tonight-awake-audiocpp-current.wav
audio.cpp output from the official
tonight-awake
example.
examples/tonight-awake-audiocpp-current_q8.wav
Same official example generated with
yue2-3b-q8_0.gguf
and
yue2-vae-f16.gguf
.
examples/yue2_demo.mp4
Short audio.cpp Yue2 WebUI demo video.
Demos
The demos below were generated with
yue2-3b-bf16.gguf
and
yue2-vae-f32.gguf
.
The
_q8.wav
comparison files were generated with
yue2-3b-q8_0.gguf
and
yue2-vae-f16.gguf
using the same prompts and seeds.
Replace
<AUDIOCPP>
with your audio.cpp checkout and
<YUE2_GGUF>
with this
downloaded GGUF repo directory.
WebUI demo video
Sorry if the UI is a bit rough. I’m not a UI person and probably messed up some of the nice community UI work while wiring this up.
<AUDIOCPP>/build/debug/bin/audiocpp_cli \
--task gen \
--family yue2 \
--model <YUE2_GGUF> \
--backend cuda \
--threads 8 \
--text $'[Verse]\nSoft morning light is touching the window.\nI hear the city waking below.\n[Chorus]\nStay with the rhythm, let it carry us home.\nSing with the sunrise, we are never alone.' \
--request-option 'style=English, indie pop, bright acoustic guitar, soft drums, warm lead vocal, polished demo mix' \
--request-option cot=off \
--request-option seed=20260920 \
--request-option num_inference_steps=8 \
--session-option yue2.model_gguf=yue2-3b-bf16.gguf \
--session-option yue2.vae_gguf=yue2-vae-f32.gguf \
--out demo1.wav \
--log
Demo 2: full planning
Q8 + F16 VAE:
<AUDIOCPP>/build/debug/bin/audiocpp_cli \
--task gen \
--family yue2 \
--model <YUE2_GGUF> \
--backend cuda \
--threads 8 \
--text $'[Verse]\nA quiet piano opens the evening.\nWarm city lights are starting to glow.\n[Chorus]\nLift every voice into the skyline.\nLet the melody carry us home.' \
--request-option 'style=English, piano pop, clear lead vocal, gentle bass, soft drums, warm chorus harmonies' \
--request-option cot=full \
--request-option seed=20260921 \
--request-option num_inference_steps=8 \
--session-option yue2.model_gguf=yue2-3b-bf16.gguf \
--session-option yue2.vae_gguf=yue2-vae-f32.gguf \
--out demo2.wav \
--log
<AUDIOCPP>/build/debug/bin/audiocpp_cli \
--task gen \
--family yue2 \
--model <YUE2_GGUF> \
--backend cuda \
--threads 8 \
--text $'[Verse]\nThe room begins with a gentle phrase.\nThe harmony turns through a golden haze.\n[Chorus]\nHold the note and let it rise.\nMorning opens up the sky.' \
--request-option 'style=English, piano pop with jazz harmony, clear lead vocal, gentle bass, tasteful drum fills' \
--request-option cot=full \
--request-option abc_file=<AUDIOCPP>/reference/YuE/examples/score-jazz.abc \
--request-option seed=20260923 \
--request-option num_inference_steps=8 \
--session-option yue2.model_gguf=yue2-3b-bf16.gguf \
--session-option yue2.vae_gguf=yue2-vae-f32.gguf \
--out demo4.wav \
--log
Upstream revision used for this package:
1a96eca688d6ae5d7f0feb88573fec89920fcd19
License
The upstream model is released under
cc-by-nc-4.0
. This GGUF package follows
the same license. See the upstream repository for the full license terms and
model card details.
Runs of audio-cpp Yue2-3B-GGUF on huggingface.co
115.4K
Total runs
0
24-hour runs
20.1K
3-day runs
42.1K
7-day runs
108.6K
30-day runs
More Information About Yue2-3B-GGUF huggingface.co Model
Yue2-3B-GGUF huggingface.co is an AI model on huggingface.co that provides Yue2-3B-GGUF's model effect (), which can be used instantly with this audio-cpp Yue2-3B-GGUF model. huggingface.co supports a free trial of the Yue2-3B-GGUF model, and also provides paid use of the Yue2-3B-GGUF. Support call Yue2-3B-GGUF model through api, including Node.js, Python, http.
Yue2-3B-GGUF huggingface.co is an online trial and call api platform, which integrates Yue2-3B-GGUF's modeling effects, including api services, and provides a free online trial of Yue2-3B-GGUF, you can try Yue2-3B-GGUF online for free by clicking the link below.
audio-cpp Yue2-3B-GGUF online free url in huggingface.co:
Yue2-3B-GGUF is an open source model from GitHub that offers a free installation service, and any user can find Yue2-3B-GGUF on GitHub to install. At the same time, huggingface.co provides the effect of Yue2-3B-GGUF install, users can directly use Yue2-3B-GGUF installed effect in huggingface.co for debugging and trial. It also supports api for free installation.