bintis/uta-studio

Uta! Studio GGUF models

GGUF model files loaded by the native GGML runtime of Uta! Studio. Uta! Studio's Runtime Manager downloads them on explicit user confirmation; each file is published unchanged from the build that Uta! Studio runs.

These files are converted or repackaged from third-party upstream models. Each model keeps its own upstream license and attribution, listed below. This repository does not relicense any model. Check the upstream terms before use, in particular for non-commercial (cc-by-nc-sa-4.0) and unresolved checkpoint licenses.

The two BS-RoFormer Leap XE90 models are not mirrored here; Uta! Studio downloads them from scragnog/HOT-Step-CPP-SuperSep.

Files

Path	Model	Purpose	Size (bytes)	SHA-256
bs_polarformer_public_instrumental/model-fp16.gguf	BS-PolarFormer Public	Experimental single-pass GuideVocals and Instrumental-residual extraction	204237408	f5e40ac0dc7487a0c2ccb247e5b948cd6f2c7aaf46a2994023606e1e800ed2c1
melband_roformer_harmony/model-fp16.gguf	MelBand-RoFormer Lead Isolation	Lead-vocal extraction with vocal residual	457008736	d463c06a1bf5d3889a2a6be58cc469f0a996155eafb91845ff5e8c139a3d64be
melband_roformer_denoise_aufr33/model-fp16.gguf	MelBand-RoFormer Denoise	44.1 kHz stereo vocal denoise	457008736	eb03fce4c5a450f88718e8a529b8adcd653618a5d32cb55275fa212a80fef33a
melband_roformer_dereverb_anvuew/model-fp16.gguf	MelBand-RoFormer Dereverb	44.1 kHz stereo vocal dereverb	457008736	f850fb2460099df356676ce37ba48875e3c75726d7a848b42d75ff6015955ac7
rmvpe/rmvpe-f32.gguf	RMVPE	Continuous F0 tracking	361625344	1b4095d1b57818f5e812b1986ea5a7d7e6d64ccd9e1b1d7b71f4091304513fd2
fcpe/fcpe-f32.gguf	FCPE	Secondary continuous-F0 disagreement expert	43309760	6f6dcf86133608191d798897adda6fb1d9d480f7990e2c11c9e255e1d8c65673
basic_pitch/basic-pitch-f32.gguf	Basic Pitch	Optional onset and activation evidence	144512	7ffc0490679300974c6acee7bfa707d991fcf6a1b6fd3e1103e2f7325e8ac544
game_1_0_3_small/game-small-f32.gguf	GAME 1.0.3 Small	Primary singing note and boundary evidence	50734944	e7f8d27e43face0e6647a622b796a670bfea1209d8e759d9375af7c8cfbf3c82
game_1_0_3_medium/game-medium-f32.gguf	GAME 1.0.3 Medium	Primary singing note and boundary evidence	199584064	a69c52a01f452c8092f1479630074592d4c3f0ef7404bb65ddd73fab01a0606e
game_1_0_3_large/game-large-f32.gguf	GAME 1.0.3 Large	Primary singing note and boundary evidence	396034784	6c83cd92a5f36f3277f1a8be3b24155a398c9b8fe14ace1d3943aebe005692b9
jbm555_cectc_80/jbm555-cectc80-f32.gguf	JBM555 CE-CTC 80	Japanese mix-and-vocal conditioned note evidence	3981024	8c9dba8320715db8817fb17a8b2f8dcea5d59dd0809309a72b00f02dbb52b10a
stars/stars-f32.gguf	STARS Chinese P1	Timed-transcript-conditioned note, technique, and style evidence	201085024	2d845732ce308b8bf89304c0c854557246a76032f593df9f0dc42c447aa9a893
rosvot/rosvot-f32.gguf	ROSVOT P0	Timed-transcript-conditioned singing note evidence	48194496	ae208457b04cc11ef2dcd063dabf2d0864785a17981955366aac888390c148c8
firered_asr2_aed/cmvn.ark	FireRedASR2-AED	Optional Mandarin, dialect, English, code-switching, and singing transcript challenger	1311	6efba6105429d1630c05d818d956bfe4edfad37a04b3b27bb5a029b9adb37945
firered_asr2_aed/dict.txt	FireRedASR2-AED	Optional Mandarin, dialect, English, code-switching, and singing transcript challenger	79172	1bc613de2112d257e61a349c3e72d1b1a9cf19c33d3ca954197ad2171e5ea07b
firered_asr2_aed/firered-f32.gguf	FireRedASR2-AED	Optional Mandarin, dialect, English, code-switching, and singing transcript challenger	4686918112	7724d4f01ac8c208670be968cef236b73f4276eddd0de8f85441b56bb6e9d132
qwen3_asr_1_7b/Qwen3-ASR-1.7B-F16.gguf	Qwen3-ASR 1.7B	Primary singing transcription	4083087904	b20587d247ae3d3b2e82f111944b7f3cb98a2a272068bf64580c191bf3b8272b
qwen3_forced_aligner_0_6b/Qwen3-ForcedAligner-0.6B-F16.gguf	Qwen3 Forced Aligner 0.6B	Word-level forced alignment	1842216416	c70553d4e363b752db9110bba0a1ef5fb87355cd80e14703c457fbe7f39a936b

Sources and licenses

BS-PolarFormer Public (bs_polarformer_public_instrumental)

• License status: mit
• Attribution: bgkb public BS-PolarFormer model
• Source page: https://huggingface.co/bgkb/bs_polarformer
• Source: bgkb/bs_polarformer @ 9158719ee2173edd480a735764627526506fe4af, model-fp16.gguf

MelBand-RoFormer Lead Isolation (melband_roformer_harmony)

• License status: informational
• Attribution: aufr33 + viperx MelBand RoFormer Karaoke
• Source page: https://github.com/TRvlvr/model_repo/releases/tag/all_public_uvr_models
• Source: https://github.com/TRvlvr/model_repo @ all_public_uvr_models, mel_band_roformer_karaoke_aufr33_viperx_sdr_10.1956.ckpt

MelBand-RoFormer Denoise (melband_roformer_denoise_aufr33)

• License status: informational
• Attribution: aufr33 MelBand RoFormer denoise
• Source page: https://huggingface.co/poiqazwsx/melband-roformer-denoise
• Source: poiqazwsx/melband-roformer-denoise @ 4e39bc34a36dda8e73254cd8f5d44f15de2bd7b9, denoise_mel_band_roformer_aufr33_sdr_27.9959.ckpt

MelBand-RoFormer Dereverb (melband_roformer_dereverb_anvuew)

• License status: informational
• Attribution: anvuew MelBand RoFormer dereverb
• Source page: https://huggingface.co/anvuew/dereverb_mel_band_roformer
• Source: anvuew/dereverb_mel_band_roformer @ cef05ad2b5b3145ea5c149d3ad5d1f8439b34d06, dereverb_mel_band_roformer_anvuew_sdr_19.1729.ckpt

RMVPE (rmvpe)

• License status: informational
• Attribution: RMVPE model

FCPE (fcpe)

• License status: mit
• Attribution: CNChTu/FCPE canonical project; gzivdo community ONNX export
• Source page: https://huggingface.co/gzivdo/fcpe-onnx/tree/5800a2b1944967f55bb0bfeb9718cb749f809310
• Source: https://huggingface.co/gzivdo/fcpe-onnx @ 5800a2b1944967f55bb0bfeb9718cb749f809310, fcpe.onnx
• Algorithm: https://github.com/CNChTu/FCPE @ 6a149c1afb1c7e7821b71869dfb31ad50c95b516 (MIT)
• Upstream artifact: fcpe.onnx sha256 b7e4f3871b10641869b7ac5a2d56ed94deb37552c0336d77e17ad6e66760adf0

Basic Pitch (basic_pitch)

• License status: apache-2.0
• Attribution: Spotify Basic Pitch; selected source bytes are from the AEmotionStudio mirror
• Source page: https://huggingface.co/AEmotionStudio/basic-pitch-onnx-models/tree/327fd8ccd2f0bb84cbe56b4a0e9d318398ddf763
• Source: https://huggingface.co/AEmotionStudio/basic-pitch-onnx-models @ 327fd8ccd2f0bb84cbe56b4a0e9d318398ddf763, basic-pitch-f32.gguf
• Algorithm: https://github.com/spotify/basic-pitch @ fa5997af0a8210982619003269994a1be25eddf3 (Apache-2.0)
• Upstream artifact: nmp.onnx sha256 2c3c1d144bfa61ad236e92e169c13535c880469a12a047d4e73451f2c059a0ec

GAME 1.0.3 Small (game_1_0_3_small)

• License status: cc-by-nc-sa-4.0
• Attribution: openvpi GAME 1.0.3 small model
• Source page: https://github.com/openvpi/GAME/releases/tag/v1.0.3
• Source: https://github.com/openvpi/GAME @ 475a8ee781fe8cca980b3b12fbe6c80c768a813a, game-small-f32.gguf
• Algorithm: https://github.com/openvpi/GAME @ 475a8ee781fe8cca980b3b12fbe6c80c768a813a (CC-BY-NC-SA-4.0)

GAME 1.0.3 Medium (game_1_0_3_medium)

• License status: cc-by-nc-sa-4.0
• Attribution: openvpi GAME 1.0.3 medium model
• Source page: https://github.com/openvpi/GAME/releases/tag/v1.0.3
• Source: https://github.com/openvpi/GAME @ 475a8ee781fe8cca980b3b12fbe6c80c768a813a, game-medium-f32.gguf
• Algorithm: https://github.com/openvpi/GAME @ 475a8ee781fe8cca980b3b12fbe6c80c768a813a (CC-BY-NC-SA-4.0)

GAME 1.0.3 Large (game_1_0_3_large)

• License status: cc-by-nc-sa-4.0
• Attribution: openvpi GAME 1.0.3 large model
• Source page: https://github.com/openvpi/GAME/releases/tag/v1.0.3
• Source: https://github.com/openvpi/GAME @ 475a8ee781fe8cca980b3b12fbe6c80c768a813a, game-large-f32.gguf
• Algorithm: https://github.com/openvpi/GAME @ 475a8ee781fe8cca980b3b12fbe6c80c768a813a (CC-BY-NC-SA-4.0)

JBM555 CE-CTC 80 (jbm555_cectc_80)

• License status: informational
• Attribution: york135 CE-CTC baseline for APSIPA 2025
• Source page: https://github.com/york135/CECTC_baseline_APSIPA25
• Source: https://github.com/york135/CECTC_baseline_APSIPA25 @ d1352eda1ea69d94cf7b1b06bf0b003d874b389a, jbm555-cectc80-f32.gguf

STARS Chinese P1 (stars)

• License status: checkpoint-license-unresolved
• Attribution: gwx314/STARS MIT source; verstar/STARS checkpoint provenance is recorded separately
• Source page: https://huggingface.co/verstar/STARS/tree/744a7ad02e1d788452293cd903ea6a933f7862c4
• Source: https://huggingface.co/verstar/STARS @ 744a7ad02e1d788452293cd903ea6a933f7862c4, stars-f32.gguf
• Algorithm: https://github.com/gwx314/STARS @ f0e43e96cfe953f71a6cf9efd8b908b2c9d7e167 (MIT)
• Upstream artifact: model_ckpt_steps_200000.ckpt sha256 9159dd37516918448b0815ed86e1e3976d39c3044117da78db0ef65d1941db3c

ROSVOT P0 (rosvot)

• License status: checkpoint-license-unresolved
• Attribution: RickyL-2000/ROSVOT MIT source; selected checkpoint provenance is recorded separately
• Source page: https://github.com/RickyL-2000/ROSVOT/tree/3c8332bf43adae35f6e4d64971862f2f6139b310
• Source: https://github.com/RickyL-2000/ROSVOT @ 3c8332bf43adae35f6e4d64971862f2f6139b310, rosvot-f32.gguf
• Algorithm: https://github.com/RickyL-2000/ROSVOT @ 3c8332bf43adae35f6e4d64971862f2f6139b310 (MIT)
• Upstream artifact: rosvot sha256 7501fb5f913d971c2f51bcb3063b930027b03206581820a4d2bfdc394c9c3fcb

FireRedASR2-AED (firered_asr2_aed)

• License status: apache-2.0
• Attribution: FireRedTeam FireRedASR2-AED canonical checkpoint
• Source page: https://huggingface.co/FireRedTeam/FireRedASR2-AED/tree/2304afed56eacfee6256dee5937ed22ffa0b64ec
• Source: https://huggingface.co/FireRedTeam/FireRedASR2-AED @ 2304afed56eacfee6256dee5937ed22ffa0b64ec, model.pth.tar
• Algorithm: https://github.com/FireRedTeam/FireRedASR2S @ 4e7d9aaf4482a47cec1724807026b9b151926eb5 (Apache-2.0)
• Upstream artifact: cmvn.ark sha256 6efba6105429d1630c05d818d956bfe4edfad37a04b3b27bb5a029b9adb37945
• Upstream artifact: dict.txt sha256 1bc613de2112d257e61a349c3e72d1b1a9cf19c33d3ca954197ad2171e5ea07b

Qwen3-ASR 1.7B (qwen3_asr_1_7b)

• License status: apache-2.0
• Attribution: Qwen canonical model weights converted locally to GGUF
• Source page: https://huggingface.co/Qwen/Qwen3-ASR-1.7B
• Source: https://huggingface.co/Qwen/Qwen3-ASR-1.7B @ 7278e1e70fe206f11671096ffdd38061171dd6e5, Qwen3-ASR-1.7B-F16.gguf

Qwen3 Forced Aligner 0.6B (qwen3_forced_aligner_0_6b)

• License status: apache-2.0
• Attribution: Qwen canonical model weights converted locally to GGUF
• Source page: https://huggingface.co/Qwen/Qwen3-ForcedAligner-0.6B-hf
• Source: https://huggingface.co/Qwen/Qwen3-ForcedAligner-0.6B-hf @ c07281df297b9905d24a508279258cccf987a064, Qwen3-ForcedAligner-0.6B-F16.gguf