MVSEP Logo
  • 主页
  • 新闻
  • 套餐
  • 示例
  • 常见问题解答
  • 创建账号
  • 登录

开始

MVSEP API 提供编程访问,用于创建分离、返回结果、获取站点信息和控制用户功能。

要使用此 API,您需要一个 API 密钥。 您尚未登录,请登录以获取您的 API 令牌。

Examples in Python: https://github.com/ZFTurbo/MVSep-API-Examples

Unofficial application MacVSep for Mac OS: https://github.com/septcoco/macvsep

Webhooks can be setup here: Webhooks Page

请注意,Premium 用户最多可申请 10 个并行任务, 而非 Premium 用户只能请求 1 个并行任务。

分离活动

创建分离任务

要创建分离任务,您需要对以下网址进行 POST 调用:
https://mvsep.com/api/separation/create


查询参数

字段 类型 说明
api_token String 您的 API Key
audiofile Binary 要处理的音频文件. Not required if url parameter is set
url String Link to the audiofile. Not required if audiofile parameter is set
remote_type String (optional) Remote type of the audiofile. Defaults to direct. Not required if audiofile parameter is set. Supported types are direct (Direct Download), mega (MEGA), drive (Google Drive), dropbox (Dropbox)
sep_type Integer (可选) 分离类型,默认为11 (Demucs3 Model B (人声,鼓声,贝斯,其他))
add_opt1 String (可选) 分离类型的第一个额外字段
add_opt2 String (可选) 分离类型的第二个额外字段
add_opt3 String (可选)分离类型第三个附加字段
output_format Integer (可选) 输出格式,默认为0 (mp3 (320kbps))
is_demo Boolean (可选) 是否将分离结果发布到示例页,默认为false

Separation type (sep_type) and additional fields (add_opt1, add_opt2, ...) values

名称 值 Additional fields
Ensemble (vocals, instrum) 26 Key: add_opt1
Name: Output files
Options:
  • 0 - Standard set (default)
  • 1 - Include intermediate results and max_fft, min_fft

Key: add_opt2
Name: Model Type
Options:
  • 1 - SDR Vocals 10.44 (MDX23C, VitLarge23, Demucs4HT)
  • 2 - SDR Vocals 10.75 (MDX23C, BS Roformer (v1), VitLarge23)
  • 3 - SDR Vocals 11.06 (MDX23C, BS Roformer (viperx))
  • 4 - SDR Vocals 11.33 (MDX23C, BS Roformer (finetuned))
  • 5 - SDR Vocals 11.50 (Mel Roformer and BS Roformer)
  • 6 - SDR Vocals 11.61 (Mel Roformer, BS Roformer and SCNet XL)
  • 7 - SDR Vocals 11.93 (Mel Roformer, BS Roformer (x2) and SCNet XL IHF) (default)
  • 8 - High Vocal Fullness (SDR: 11.69, Fullness: 20.46)
  • 9 - High Instrumental Fullness (SDR: 17.69, Fullness: 34.79)
Ensemble (vocals, instrum, bass, drums, other) 28 Key: add_opt1
Name: Output files
Options:
  • 0 - Standard set (default)
  • 1 - Include intermediate results and max_fft, min_fft

Key: add_opt2
Name: Model Type
Options:
  • 1 - SDR average: 11.21 (v. 2023.09.01)
  • 2 - SDR average: 11.87 (v. 2024.03.08)
  • 3 - SDR average: 12.03 (v. 2024.03.28)
  • 4 - SDR average: 12.17 (v. 2024.04.04)
  • 5 - SDR average: 12.34 (v. 2024.05.21)
  • 6 - SDR average: 12.66 (v. 2024.07.14)
  • 7 - SDR average: 12.76 (v. 2024.08.15)
  • 8 - SDR average: 12.84 (v. 2024.10.08)
  • 9 - SDR average: 13.01 (v. 2024.12.20)
  • 10 - SDR average: 13.07 (v. 2024.12.28)
  • 11 - SDR average: 13.67 (v. 2025.06.30) (default)
Ensemble All-In (vocals, bass, drums, piano, guitar, lead/back vocals, other) 30 Key: add_opt1
Name: Output files
Options:
  • 0 - Standard set (default)
  • 1 - Include intermediate results and max_fft, min_fft

Key: add_opt2
Name: Model Type
Options:
  • 1 - SDR average: 11.21 (v. 2023.09.01)
  • 2 - SDR average: 11.87 (v. 2024.03.08)
  • 3 - SDR average: 12.03 (v. 2024.03.28)
  • 4 - SDR average: 12.17 (v. 2024.04.04)
  • 5 - SDR average: 12.32 (v. 2024.05.21)
  • 6 - SDR average: 12.66 (v. 2024.07.14)
  • 7 - SDR average: 12.76 (v. 2024.08.15)
  • 8 - SDR average: 12.84 (v. 2024.10.08)
  • 9 - SDR average: 13.01 (v. 2024.12.20)
  • 10 - SDR average: 13.07 (v. 2024.12.28)
  • 11 - SDR average: 13.67 (v. 2025.06.30) (default)
BS Roformer SW (vocals, bass, drums, guitar, piano, other) 63
BS Roformer (vocals, instrumental) 40 Key: add_opt1
Name: Vocal model type
Options:
  • 3 - ver. 2024.02 (SDR vocals: 10.42, SDR instrum: 16.73)
  • 4 - viperx edition (SDR vocals: 10.87, SDR instrum: 17.17)
  • 5 - ver 2024.04 (SDR vocals: 11.24, SDR instrum: 17.55)
  • 29 - ver 2024.08 (SDR vocals: 11.31, SDR instrum: 17.62)
  • 85 - unwa high instrum fullness (SDR instrum: 17.25)
  • 81 - ver 2025.07 (SDR vocals: 11.89, SDR instrum: 18.20) (default)
MelBand Roformer (vocals, instrumental) 48 Key: add_opt1
Name: Vocal model type
Options:
  • 0 - Kimberley Jensen edition (SDR vocals: 11.01, SDR instrum: 17.32)
  • 1 - ver 2024.08 (SDR vocals: 11.17, SDR instrum: 17.48)
  • 2 - Bas Curtiz edition (SDR vocals: 11.18, SDR instrum: 17.49)
  • 3 - unwa Instrumental v1 (SDR vocals: 10.24, SDR instrum: 16.54)
  • 5 - unwa Instrumental v1e (SDR vocals: 10.05, SDR instrum: 16.36)
  • 6 - unwa big beta v5e (SDR vocals: 10.59, SDR instrum: 16.89)
  • 4 - ver 2024.10 (SDR vocals: 11.28, SDR instrum: 17.59) (default)
  • 7 - becruily instrum high fullness (SDR instrum: 16.47)
  • 8 - becruily vocals high fullness (SDR vocals: 10.55)
  • 9 - unwa Instrumental v1e plus (SDR vocals: 10.33, SDR instrum: 16.64)
  • 10 - gabox Instrumental v7 (SDR vocals: 10.32, SDR instrum: 16.63)
MDX23C (vocals, instrumental) 25 Key: add_opt1
Name: Vocal model type
Options:
  • 3 - 12K FFT, Large Conv, Hop 1024 (SDR vocals: 9.95, SDR instrum: 16.26)
  • 2 - 12K FFT, Large Conv (SDR vocals: 9.71, SDR instrum: 16.02)
  • 0 - 12K FFT (SDR vocals: 9.68, SDR instrum: 15.99)
  • 1 - 12K FFT, 6 Poolings (SDR vocals: 9.49, SDR instrum: 15.79)
  • 4 - 8K FFT (SDR vocals: 10.17, SDR instrum: 16.48)
  • 7 - 8K FFT (SDR vocals: 10.36, SDR instrum: 16.66) (default)
SCNet (vocals, instrumental) 46 Key: add_opt1
Name: Vocal model type
Options:
  • 0 - SCNet (SDR vocals: 10.25, SDR instrum: 16.56)
  • 1 - SCNet Large (SDR vocals: 10.74, SDR instrum: 17.05)
  • 2 - SCNet XL (SDR vocals: 10.96, SDR instrum: 17.27)
  • 3 - SCNet XL (high fullness)
  • 4 - SCNet XL (very high fullness)
  • 5 - SCNet XL IHF (SDR vocals: 11.11, SDR instrum: 17.41) (default)
Demucs4 HT (vocals, drums, bass, other) 20 Key: add_opt1
Name: Model type
Options:
  • 0 - htdemucs_ft (High Quality, Slow)
  • 1 - htdemucs (Good Quality, Fast) (default)
  • 2 - htdemucs_6s (6 stems, additional piano and guitar)
MDX B (vocals, instrumental) 23 Key: add_opt1
Name: Vocal model type
Options:
  • 7 - MDX Kimberley Jensen v2 2023.05.21 (SDR: 9.60)
  • 0 - MDX UVR 2022.01.01 (SDR: 8.83)
  • 1 - MDX UVR 2022.07.25(SDR: 8.67)
  • 2 - MDX Kimberley Jensen v1 2023.02.12 (SDR: 9.48)
  • 4 - UVR-MDX-NET-Inst_HQ_2 (SDR: 9.12)
  • 5 - UVR_MDXNET_Main (SDR: 8.79)
  • 6 - MDX Kimberley Jensen Inst (SDR: 9.28)
  • 8 - UVR-MDX-NET-Inst_HQ_3 (SDR: 9.38)
  • 9 - UVR-MDX-NET-Voc_FT (SDR: 9.64)
  • 11 - UVR-MDX-NET-Inst_HQ_4 (SDR: 9.71) (default)
  • 12 - UVR-MDX-NET-Inst_HQ_5 (SDR: 9.45)
Ultimate Vocal Remover VR (vocals, music) 9 Key: add_opt1
Name: Model Type
Options:
  • 0 - HP2-4BAND-3090_4band_arch-500m_1 (default)
  • 1 - HP2-4BAND-3090_4band_2
  • 2 - HP2-4BAND-3090_4band_1
  • 3 - HP_4BAND_3090
  • 4 - Vocal_HP_4BAND_3090
  • 5 - Vocal_HP_4BAND_3090_AGG
  • 6 - HP2-MAIN-MSB2-3BAND-3090
  • 7 - HP-4BAND-V2
  • 8 - HP-KAROKEE-MSB2-3BAND-3090 (Karaokee model)
  • 9 - WIP-Piano-4band-129605kb (Piano model)
  • 10 - drums-4BAND-3090_4band (Drums model)
  • 11 - bass-4BAND-3090_4band (Bass model)
  • 12 - karokee_4band_v2_sn (Karaokee model v2)
  • 13 - UVR-De-Echo-Aggressive
  • 14 - UVR-De-Echo-Normal
  • 15 - UVR-DeNoise
  • 16 - UVR-DeEcho-DeReverb
  • 17 - UVR-BVE-4B_SN-44100-1 (Back vocals model)

Key: add_opt2
Name: Agressiveness
Options:
  • 0.3 - 0.3 (default)
  • 0.1 - 0.1
  • 0.2 - 0.2
  • 0.4 - 0.4
  • 0.5 - 0.5
  • 0.6 - 0.6
  • 0.7 - 0.7
  • 0.8 - 0.8
  • 0.9 - 0.9
  • 1.0 - 1.0
Demucs4 Vocals 2023 (vocals, instrum) 27
MelBand Karaoke (lead/back vocals) 49 Key: add_opt1
Name: Karaoke model type
Options:
  • 0 - Model by viperx and aufr33 (SDR: 9.45)
  • 1 - Model by becruily (SDR: 9.61)
  • 2 - Model by gabox (SDR: 9.67)
  • 3 - Model fuzed gabox & aufr33/viperx (SDR: 9.85) (default)

Key: add_opt2
Name: Extraction type
Options:
  • 0 - Use as is (default)
  • 1 - Extract vocals first
MDX-B Karaoke (lead/back vocals) 12 Key: add_opt1
Name: Karaoke model type
Options:
  • 0 - Extract directly from mixture (SDR lead vocals: 6.81) (default)
  • 1 - Extract from vocals part (SDR lead vocals: 7.94)
MVSep Piano (piano, other) 29 Key: add_opt1
Name: Piano model type
Options:
  • 0 - mdx23c (2023.08, SDR: 4.79)
  • 1 - mdx23c (2024.09, SDR: 5.59)
  • 2 - MelRoformer (viperx, SDR: 5.71)
  • 3 - SCNet Large (2024.09, SDR: 5.89)
  • 4 - Ensemble (SCNet + Mel, SDR: 6.20)
  • 5 - BS Roformer SW (SDR: 7.83) (default)
MVSep Guitar (guitar, other) 31 Key: add_opt1
Name: Guitar model type
Options:
  • 0 - mdx23c (2023.08, SDR: 4.78)
  • 2 - mdx23c (2024.06, SDR: 6.34)
  • 3 - MelRoformer (2024.06, SDR: 7.02)
  • 5 - BSRoformer (viperx, SDR: 7.16)
  • 6 - Ensemble (BS + Mel, SDR: 7.51)
  • 7 - BS Roformer SW (SDR: 9.05) (default)
MVSep Bass (bass, other) 41 Key: add_opt1
Name: Bass model type
Options:
  • 0 - BS Roformer (SDR bass: 12.49)
  • 1 - HTDemucs4 (SDR bass: 12.52)
  • 2 - SCNet XL (SDR bass: 13.81)
  • 3 - BS + HTDemucs + SCNet (SDR bass: 14.07)
  • 4 - BS Roformer SW (SDR bass: 14.62)
  • 5 - BS Roformer SW + SCNet XL (SDR bass: 14.87) (default)

Key: add_opt2
Name: How to extract
Options:
  • 0 - Extract directly from mixture (default)
  • 1 - Extract from instrumental part

Key: add_opt3
Name: Output files
Options:
  • 0 - Standard set (default)
  • 1 - Include results of independent models
MVSep Drums (drums, other) 44 Key: add_opt1
Name: Drums model type
Options:
  • 0 - HTDemucs (SDR drums: 12.04)
  • 1 - MelBand Roformer (SDR drums: 12.76)
  • 2 - SCNet Large (SDR drums: 13.01)
  • 3 - SCNet XL (SDR drums: 13.42)
  • 4 - Mel + SCNet XL (SDR drums: 13.78)
  • 5 - BS Roformer SW (SDR drums: 14.11)
  • 6 - Mel + SCNet XL + BS Roformer SW (SDR drums: 14.35) (default)

Key: add_opt2
Name: How to extract
Options:
  • 0 - Extract directly from mixture (default)
  • 1 - Extract from instrumental part

Key: add_opt3
Name: Output files
Options:
  • 0 - Standard set (default)
  • 1 - Include results of independent models
MVSep Strings (strings, other) 52
MVSep Wind (wind, other) 54 Key: add_opt1
Name: Wind model type
Options:
  • 0 - MelBand Roformer (SDR wind: 6.73)
  • 1 - SCNet Large (SDR wind: 6.76)
  • 2 - Mel + SCNet (SDR wind: 7.22) (default)

Key: add_opt2
Name: How to extract
Options:
  • 0 - Extract directly from mixture (default)
  • 1 - Extract from instrumental part

Key: add_opt3
Name: Output files
Options:
  • 0 - Standard set (default)
  • 1 - Include results of independent models
MVSep Organ (organ, other) 58 Key: add_opt1
Name: Organ model type
Options:
  • 0 - SCNet XL (SDR organ: 2.71)
  • 1 - MelBand Roformer (SDR organ: 2.77)
  • 2 - Mel + SCNet (SDR organ: 3.05) (default)
MVSep Saxophone (saxophone, other) 61 Key: add_opt1
Name: Model type
Options:
  • 0 - SCNet XL (SDR saxophone: 6.15)
  • 1 - MelBand Roformer (SDR saxophone: 6.97) (default)
  • 2 - Mel + SCNet (SDR saxophone: 7.13)
MVSep Violin (violin, other) 65
Apollo Enhancers (by JusperLee and Lew) 51 Key: add_opt1
Name: Model type
Options:
  • 0 - MP3 Enhancer (by JusperLee) (default)
  • 1 - Universal Super Resolution (by Lew)
  • 2 - Vocals Super Resolution (by Lew)
  • 3 - Universal Super Resolution (by MVSep Team)
Reverb Removal (noreverb) 22 Key: add_opt1
Name: Model Type
Options:
  • 0 - Reverb removal by FoxJoy (MDX23C) (default)
  • 1 - Reverb removal by anvuew (MelRoformer)
  • 2 - Reverb removal by anvuew (BSRoformer)
  • 3 - Reverb removal by anvuew v2 (MelRoformer)
  • 4 - Reverb removal by Sucial (MelRoformer)
  • 5 - Reverb removal by Sucial v2 (MelRoformer)

Key: add_opt2
Name: Preprocess
Options:
  • 0 - Extract vocals (needed for Mel/BS Roformer)
  • 1 - Use as is (default)
MVSep Crowd removal (crowd, other) 34 Key: add_opt1
Name: Model type
Options:
  • 8 - MDX23C v1 (SDR crowd: 5.57)
  • 9 - MDX23C v2 (SDR crowd: 6.06)
  • 0 - Mel Roformer (SDR crowd: 6.07)
  • 1 - Ensemble MDX23C + Mel Roformer (SDR crowd: 6.27) (default)
MVSep Demucs4HT DNR (dialog, sfx, music) 24 Key: add_opt1
Name: Model type
Options:
  • 0 - Single (SDR: 9.62) (default)
  • 1 - Ensemble (SDR: 10.16)
BandIt Plus (speech, music, effects) 36
BandIt v2 (speech, music, effects) 45 Key: add_opt1
Name: Model Type
Options:
  • 0 - Multi language model (default)
  • 1 - English model
  • 2 - German model
  • 3 - French model
  • 4 - Spanish model
  • 5 - Chinese model
  • 6 - Faroese model
MVSep DnR v3 (speech, music, sfx) 56 Key: add_opt1
Name: Model type
Options:
  • 0 - SCNet Large (SDR avg: 11.22)
  • 1 - MelBand Roformer (SDR avg: 10.99)
  • 2 - Mel + SCNet (SDR avg: 11.54) (default)

Key: add_opt2
Name: How to extract
Options:
  • 0 - Extract directly from mixture (default)
  • 1 - Use vocals model to help

Key: add_opt3
Name: Output files
Options:
  • 0 - Standard set (default)
  • 1 - Include results of independent models
DrumSep (4-6 stems: kick, snare, cymbals, toms, ride, hh, crash) 37 Key: add_opt1
Name: Model Type
Options:
  • 0 - DrumSep model by inagoy (HDemucs, 4 stems)
  • 1 - DrumSep model by aufr33 and jarredou (MDX23C, 4 stems)
  • 2 - DrumSep SCNet XL (5 stems)
  • 3 - DrumSep SCNet XL (6 stems)
  • 4 - DrumSep SCNet XL (4 stems)
  • 5 - DrumSep Ensemble of 4 models (MDX23C + 3 * SCNet XL, 8 stems)
  • 6 - DrumSep MelBand Roformer (4 stems)
  • 7 - DrumSep MelBand Roformer (6 stems) (default)

Key: add_opt2
Name: Preprocess
Options:
  • 0 - Apply Drums model before (default)
  • 1 - Use as is (audio must contain drums only)
DeNoise by aufr33 47 Key: add_opt1
Name: Model type
Options:
  • 0 - Standard (default)
  • 1 - Aggressive
Whisper (extract text from audio) 39 Key: add_opt1
Name: Model type
Options:
  • 0 - Apply to original file
  • 1 - Extract vocals first (default)

Key: add_opt2
Name: Transcription type
Options:
  • 0 - New timestamps by linto-ai
  • 1 - Old version of timestamps by whisper (default)
Parakeet (extract text from audio) 64 Key: add_opt1
Name: Model type
Options:
  • 0 - Apply to original file
  • 1 - Extract vocals first (default)
Medley Vox (Multi-singer separation) 53 Key: add_opt1
Name: Model type
Options:
  • 0 - Apply to original file
  • 1 - Extract vocals first (default)
MVSep Multichannel BS (vocals, instrumental) 43 Key: add_opt1
Name: Model Type
Options:
  • 0 - BS Roformer (SDR: 11.81) (default)
  • 1 - MDX23C (SDR: 10.36)
  • 2 - MelBand Roformer (SDR: 11.17)
  • 3 - MelBand Roformer XL (SDR: 11.28)
MVSep Male/Female separation 57 Key: add_opt1
Name: Model type
Options:
  • 0 - BSRoformer by Sucial (SDR: 6.52)
  • 3 - BSRoformer by aufr33 (SDR: 8.18)
  • 1 - SCNet XL (SDR: 11.83)
  • 2 - MelRoformer (2025.01) (SDR: 13.03) (default)

Key: add_opt2
Name: How to extract
Options:
  • 0 - Extract directly from mixture (default)
  • 1 - Extract vocals first with BS Roformer
MDX A/B (vocals, drums, bass, other) 7 Key: add_opt1
Name: Vocal model type
Options:
  • 0 - MDX A (Contest Version)
  • 3 - MDX Kimberley Jensen 2023.02.12 SDR: 9.30 (New) (default)
  • 1 - MDX UVR 2022.01.01 SDR 8.62
  • 2 - MDX UVR 2022.07.25 SDR 8.51
Demucs3 Model (vocals, drums, bass, other) 10 Key: add_opt1
Name: Model type
Options:
  • 0 - Demucs3 Model A (Contest Version)
  • 1 - Demucs3 Model B (High Quality) (default)
Vit Large 23 (vocals, instrum) 33 Key: add_opt1
Name: Model type
Options:
  • 0 - v1 (SDR vocals: 9.78)
  • 1 - v2 (SDR vocals: 9.90) (default)
UVRv5 Demucs (vocals, music) 17 Key: add_opt1
Name: Model Type
Options:
  • 0 - UVR_Demucs_Model_1
  • 1 - UVR_Demucs_Model_2
  • 2 - UVR_Demucs_Model_Bag
MVSep DNR (music, sfx, speech) 18
MVSep Vocal Model (vocals, music) 19
Demucs2 (vocals, drums, bass, other) 13
Danna Sep (vocals, drums, bass, other) 15
Byte Dance (vocals, drums, bass, other) 16
MVSep MelBand Roformer (vocals, instrum) 35
spleeter 0 Key: add_opt1
Name: Model type
Options:
  • 0 - 2 stems (vocals, music) (default)
  • 1 - 4 stems (vocals, drums, bass, other)
  • 2 - 5 stems (vocals, drums, bass, piano, other)
UnMix 3 Key: add_opt1
Name: Model type
Options:
  • 0 - unmix XL (vocals, drums, bass, other) (default)
  • 1 - unmix HQ (vocals, drums, bass, other)
  • 2 - unmix SD (vocals, drums, bass, other)
  • 3 - unmix SE (vocals, music) - low quality
Zero Shot (Query Based) (Low quality) 14 Key: add_opt1
Name: Model Type
Options:
  • 0 - Bass (MUSDB18HQ AVG)
  • 1 - Drums (MUSDB18HQ AVG)
  • 2 - Vocals (MUSDB18HQ AVG) (default)
  • 3 - Other (MUSDB18HQ AVG)
LarsNet (kick, snare, cymbals, toms, hihat) 38 Key: add_opt1
Name: Model type
Options:
  • 0 - Apply Demucs4HT first to get drums (default)
  • 1 - Use as is (audio must contain drums only)
Stable Audio Open Gen 62 Key: add_opt1
Name: Text prompt
Options:

Key: add_opt2
Name: Length (in seconds)
Options:
  • 3 - 3
  • 5 - 5
  • 8 - 8
  • 10 - 10
  • 12 - 12
  • 15 - 15
  • 20 - 20
  • 25 - 25
  • 30 - 30
  • 35 - 35
  • 40 - 40
  • 45 - 45
  • 47 - 47 (default)
MVSep MultiSpeaker (MDX23C) 42 Key: add_opt1
Name: Model Type
Options:
  • 0 - Extract directly from mixture
  • 1 - Extract from vocals part (default)
Aspiration (by Sucial) 50 Key: add_opt1
Name: Model type
Options:
  • 0 - Extract directly from mixture (default)
  • 1 - Extract from vocals part
AudioSR (Super Resolution) 59 Key: add_opt1
Name: Cutoff (Hz)
Options:
  • 0 - Automatic (default)
  • 2000 - 2000
  • 3000 - 3000
  • 4000 - 4000
  • 5000 - 5000
  • 6000 - 6000
  • 7000 - 7000
  • 8000 - 8000
  • 9000 - 9000
  • 10000 - 10000
  • 11000 - 11000
  • 12000 - 12000
  • 13000 - 13000
  • 14000 - 14000
  • 15000 - 15000
  • 16000 - 16000
  • 17000 - 17000
  • 18000 - 18000
  • 19000 - 19000
  • 20000 - 20000
  • 21000 - 21000
  • 22000 - 22000
Phantom Centre extraction (by wesleyr36) 55
FlashSR (Super Resolution) 60

输出格式 (output_format) 值

名称 值
mp3 (320 kbps) 0
wav (uncompressed, 16 bit) 1
flac (lossless, 16 bit) 2
m4a (lossy) 3
wav (uncompressed, 32 bit) 4
flac (lossless, 24 bit) 5

Curl示例代码:

curl --location --request POST 'https://mvsep.com/api/separation/create' --form 'audiofile=@"/path/to/file.mp3"' --form 'api_token="<您的 API 令牌>"' --form 'sep_type="9"' --form 'add_opt1="0"' --form 'add_opt2="1"' --form 'output_format="1"' --form 'is_demo="1"'

响应详情:

键 值
success
  • false - 当创建作业失败时
  • true - 成功创建作业后
data 根据 success 键保存额外信息
data -> link 显示已创建作业的 “获取结果” URL 的链接(仅在 success 为 true 时显示)
data -> hash 显示已创建的任务哈希值(仅在 success 为 true 时显示)
data -> message 显示错误描述(仅在success为false时显示)

错误

MVSEP API 使用以下错误代码:

错误代码 含义
400 有些参数未知或无效。当您没有传入每个必需的参数,或传入了无效参数时,会出现这个错误。
401 未知或无效的api_token。如果您使用了未知的API密钥,则会出现此错误。

获取结果

要获取分离结果,需要对以下网址进行 GET 调用:
https://mvsep.com/api/separation/get


查询参数

字段 类型 说明
hash String 此分离的哈希值
mirror Integer (可选)如果设置为 1,则使用另一个镜像,默认为 0。需要 api_token 和 1 个积分(如果设置为 1)。
api_token string (可选)用户的 API 密钥。如果 mirror 参数设置为 1,则为必需。

Curl示例代码:

curl --location --request GET 'https://mvsep.com/api/separation/get?hash=20230327071601-0e3e5c6c85-13-dimensions.mp3'

响应详情:

键 值
success
  • false - 当文件哈希未找到、被删除或已过期
  • true - 当文件哈希值找到且有效
status
  • not_found - 当作业无效时
  • waiting - 当作业处于队列中且尚未处理时
  • processing - 当作业正在处理时
  • done - 当作业已成功处理
  • failed - 当作业处理失败时
  • distributing - 当作业中的音频文件较大且正在分发到多个 GPU 实例时
  • merging - 当所有 Job 分配部分完成处理并合并时
data 根据 status 键保存额外信息
data -> queue_count 在用户的优先级中,显示未处理的作业计数(仅在 status 为 waiting 或 distributing 时显示)
data -> current_order 显示用户作业的顺序(仅在status为wating或distributing时显示)
data -> message 显示状态描述(显示在所有状态上,当 status 为 failed) 时显示错误原因)
data -> algorithm 显示作业中使用的算法(仅在 status 为 done 时显示)
data -> algorithm_description 显示作业中使用的算法详细信息(仅在 status 为 done 时显示)
data -> output_format 显示作业的输出格式(仅在 status 为 done 时显示)
data -> tags 显示音频元标记(仅在 status 为 done 时显示)
data -> input_file 显示输入音频下载详情(仅当 status 为 done 时显示)
data -> files 显示输出音频下载详情(仅在 status 为 done 时显示)
data -> date 显示作业处理日期(仅在 status 为 done 时显示)
data -> finished_chunks 显示大型作业的已完成部分的数量(仅在 status 为 distributing 时显示)
data -> all_chunks 显示大型作业的所有部分的编号(仅在 status 为 distributing 时显示)

获取结果

To get the result of remote job, you need to make a GET call to the following url:
https://mvsep.com/api/separation/get-remote


查询参数

字段 类型 说明
hash String 此分离的哈希值

Curl示例代码:

curl --location --request GET 'https://mvsep.com/api/separation/get-remote?hash=n2AwBjy4KsFZfqqR'

响应详情:

键 值
success
  • false - 当文件哈希未找到、被删除或已过期
  • true - 当文件哈希值找到且有效
status
  • not_found - 当作业无效时
  • waiting - 当作业处于队列中且尚未处理时
  • processing - 当作业正在处理时
  • done - 当作业已成功处理
  • failed - 当作业处理失败时
data 根据 status 键保存额外信息
data -> queue_count Shows unprocessed jobs count in user's priority (only shown when status is waiting)
data -> current_order Shows the order of user's job (only shown when status is waiting)
data -> message Shows the description of status (only when status is failed)
data -> link Shows link to "GET RESULT" URL of created job (only shown when status is done)
data -> hash Hash of the separation (only shown when status is done)

Cancel SEPARATION

To cancel the separation and refund used credits (if the separation is not being processed yet), you need to make a POST call to the following url:
https://mvsep.com/api/separation/cancel


查询参数

字段 类型 说明
api_token string 用户 API 密钥
hash string 此分离的哈希值

Curl示例代码:

curl --location --request POST 'https://mvsep.com/api/separation/cancel' --form 'api_token="<您的 API 令牌>"' --form 'hash="<hash>"'

响应详情:

键 值
success
  • false - 当文件哈希未找到、被删除或已过期
  • true - 当文件哈希值找到且有效

站点信息

获取分离类型

要获取有关分离类型的详细数据,您需要对以下 URL 进行 GET 调用:
https://mvsep.com/api/app/algorithms


Curl示例代码:

curl --location --request GET 'https://mvsep.com/api/app/algorithms'

响应详情:

以数组形式,返回数据:

键 值
render_id 分离类型值(ID)
name 分离类型的名称
algorithm_group_id 该组属于分离类型:
  • 0 - Experimental
  • 1 - Old Models
  • 2 - HQ Models
  • 3 - Ensemble
orientation 此分离类型的对象:
  • 0 - 所有用户
  • 1 - 已注册用户
  • 2 - 高级用户
order_id 分离类型的顺序
price_coefficient 用于计算工作从高级用户那里消耗了多少积分。
使用此公式 : floor(<音频文件时长(秒)> * price_coefficient / 60)?:1
usage Shows monthly usage count of separation type
rating Shows monthly usage rating of separation type as array:
  • average - average rating
  • total - vote count
algorithm_fields 包含有关分离类型附加字段的信息
algorithm_fields -> name 显示分离类型的附加字段:
  • add_opt1 - 第一
  • add_opt2 - 第二
  • add_opt3 - 第三
您需要使用此名称进行分离,以便选择正确的附加字段。
algorithm_fields -> text 显示分离类型附加字段的名称。
algorithm_fields -> options JSON 编码的附加字段键值链
algorithm_fields -> default_key 附加字段的默认值。<code 的键之一>algorithm_fields -> options</code>
algorithm_descriptions 包含有关分离类型描述的信息
algorithm_descriptions -> short_description 不使用 HTML 的小型分离类型描述
algorithm_descriptions -> long_description 详细描述使用 HTML 的分隔类型
algorithm_descriptions -> lang 指定描述的语言代码。

获取新闻

获取关于 MVSEP 最新的信息,您需要向以下 URL 发起 GET 请求:
https://mvsep.com/api/app/news


查询参数

字段 类型 说明
lang String (可选)2 个字母的语言代码(en, ru)
start Integer (可选)起始点偏移量,默认为 0(即从最新开始)
limit Integer (可选)返回的最大结果数量,默认为 10,最大为 20。

Curl示例代码:

curl --location --request GET 'https://mvsep.com/api/app/news?lang=en&start=2&limit=1'

响应详情:

以数组形式,返回数据:

键 值
title 新闻标题
lang 新闻标题
text 新闻的完整描述
created_at 新闻日期

GET QUEUE

To get site queue information, you need to make a GET call to the following url:
https://mvsep.com/api/app/queue


查询参数

字段 类型 说明
api_token String (optional) 用户 API 密钥

Curl示例代码:

curl --location --request GET 'https://mvsep.com/api/app/queue'

响应详情:

键 值
queue Contains array of current queues for different plans:
  • in_process - Currently being processed
  • premium - Queue count for premium users
  • registered - Queue count for registered users
  • unregistered - Queue count for unregistered users
plan Contains current queue for current plan (add api_token to request for registered/premium plan):
  • plan - Your current plan
  • queue - Queue count for your current plan

获取分离演示

获取关于演示分离的信息,您需要向以下 URL 发起 GET 请求: https://mvsep.com/api/app/demo


查询参数

字段 类型 说明
start Integer (可选)起始点偏移量,默认为 0(即从最新开始)
limit Integer (可选)返回的最大结果数量,默认为 10,最大为 20。

Curl示例代码:

curl --location --request GET 'https://mvsep.com/api/app/demo?start=0&limit=2'

响应详情:

以数组形式,返回数据:

键 值
hash 此分离的哈希值
date 分离日期
input_audio 输入音频 URL
size 输入音频大小
algorithm 包含有关已用分离类型的简要信息(有关返回参数的信息,请参阅“获取分离类型”选项)

用户认证

登录

您需要向以下 URL 发送 POST 请求以验证用户并获取用户的 API 密钥:
https://mvsep.com/api/app/login

查询参数

字段 类型 说明
email String 用户电子邮件
password String 用户密码

Curl示例代码:

curl --location --request POST 'https://mvsep.com/api/app/login' --form 'email="<用户电子邮件>"' --form 'password="<用户密码>"'

响应详情:

键 值
success
  • false - 发生时指定了错误的凭据
  • true - 发生时指定了有效的凭据
data 包含凭证有效时的信息
data -> name 用户名称
data -> email 用户电子邮件
data -> api_token 用户 API 密钥。请确保将其存储在安全存储中,并用于后续操作。
data -> premium_minutes 用户剩余积分
data -> premium_enabled 用户是否允许花费积分。
  • 0 - 不允许
  • 1 - 允许
data -> long_filenames_enabled 用户是否希望在文件名中包含额外信息:
  • 0 - 较短
  • 1 - 较长

错误

MVSEP API 使用以下错误代码:

错误代码 含义
400 指定的凭证不正确

注册

注册用户到 MVSEP 网站,您需要向以下 URL 发送 POST 请求:
https://mvsep.com/api/app/register


查询参数

字段 类型 说明
name String 用户名称
email String 用户电子邮件
password String 用户密码
password_confirmation String 重复用户密码

Curl示例代码:

curl --location --request GET 'https://mvsep.com/api/app/register' --form 'name="user"' --form 'email="asdf@example.com"' --form 'password="ASDFQWEr123"' --form 'password_confirmation="ASDFQWEr123"'

响应详情:

键 值
success
  • false - 发生当电子邮件无效、用户存在或确认不正确时
  • true - 发生当表单有效时
message 发生当表单有效时

错误

MVSEP API 使用以下错误代码:

错误代码 含义
400 表单无效

配置

获取用户

获取用户信息,您需要向以下 URL 发起 GET 请求:
https://mvsep.com/api/app/user

查询参数

字段 类型 说明
api_token String 用户 API 密钥

Curl示例代码:

curl --location --request GET 'https://mvsep.com/api/app/user?api_token=<您的 API 令牌>'

响应详情:

键 值
success
  • false - 发生时指定了无效的 API 密钥
  • true - 当指定有效的 API 密钥时发生
data 包含凭证有效时的信息
data -> name 用户名称
data -> email 用户电子邮件
data -> api_token 用户 API 密钥。请确保将其存储在安全存储中,并用于后续操作。
data -> premium_minutes 用户剩余积分
data -> premium_enabled 用户是否允许花费积分。
  • 0 - 不允许
  • 1 - 允许
data -> long_filenames_enabled 用户是否希望在文件名中包含额外信息:
  • 0 - 较短
  • 1 - 较长
data -> current_queue 显示用户的持续分离

错误

MVSEP API 使用以下错误代码:

错误代码 含义
400 无效的 API 密钥

获取分离历史

获取用户基本信息分离历史,您需要向以下 URL 发起 GET 请求:
https://mvsep.com/api/app/separation_history


查询参数

字段 类型 说明
api_token String 用户 API 密钥
start Integer (可选)起始点偏移量,默认为 0(即从最新开始)
limit Integer (可选)返回的最大结果数量,默认为 10,最大为 20。

Curl示例代码:

curl --location --request GET 'https://mvsep.com/api/app/separation_history?api_token=<您的 API 令牌>&start=1&limit=2'

响应详情:

键 值
success
  • false - 发生时指定了无效的 API 密钥
  • true - 当指定有效的 API 密钥时发生
data 包含关于先前分离的基本信息数组
data -> hash 此分离的哈希值
data -> job_exists 包含关于先前分离的基本信息数组
data -> credits 作业使用了多少积分,如果分隔不存在,则为 NULL
data -> time_left 作业被删除前的预估时间
data -> algorithm 包含有关已用分离类型的简要信息(有关返回参数的信息,请参阅“获取分离类型”选项)

错误

MVSEP API 使用以下错误代码:

错误代码 含义
400 无效的 API 密钥

启用高级功能

为启用用户的高级功能,您需要向以下 URL 发送 POST 请求:
https://mvsep.com/api/app/enable_premium


查询参数

字段 类型 说明
api_token String 用户 API 密钥

Curl示例代码:

curl --location --request POST 'https://mvsep.com/api/app/enable_premium' --form 'api_token="<您的 API 令牌>"'

响应详情:

键 值
success
  • false - 发生时指定了无效的 API 密钥
  • true - 当指定有效的 API 密钥时发生
message 通知操作状态

错误

MVSEP API 使用以下错误代码:

错误代码 含义
400 无效的 API 密钥

禁用高级功能

要禁用用户的高级功能,您需要向以下 URL 发送 POST 请求:
https://mvsep.com/api/app/disable_premium


查询参数

字段 类型 说明
api_token String 用户 API 密钥

Curl示例代码:

curl --location --request POST 'https://mvsep.com/api/app/disable_premium' --form 'api_token="<您的 API 令牌>"'

响应详情:

键 值
success
  • false - 发生时指定了无效的 API 密钥
  • true - 当指定有效的 API 密钥时发生
message 通知操作状态

错误

MVSEP API 使用以下错误代码:

错误代码 含义
400 无效的 API 密钥

启用长文件名

为了启用用户的长文件名,您需要向以下 URL 发送 POST 请求:
https://mvsep.com/api/app/enable_long_filenames


查询参数

字段 类型 说明
api_token String 用户 API 密钥

Curl示例代码:

curl --location --request POST 'https://mvsep.com/api/app/enable_long_filenames' --form 'api_token="<您的 API 令牌>"'

响应详情:

键 值
success
  • false - 发生时指定了无效的 API 密钥
  • true - 当指定有效的 API 密钥时发生
message 通知操作状态

错误

MVSEP API 使用以下错误代码:

错误代码 含义
400 无效的 API 密钥

禁用长文件名

要禁用用户的长文件名,您需要向以下 URL 发送 POST 请求:
https://mvsep.com/api/app/disable_long_filenames


查询参数

字段 类型 说明
api_token String 用户 API 密钥

Curl示例代码:

curl --location --request POST 'https://mvsep.com/api/app/disable_long_filenames' --form 'api_token="<您的 API 令牌>"'

响应详情:

键 值
success
  • false - 发生时指定了无效的 API 密钥
  • true - 当指定有效的 API 密钥时发生
message 通知操作状态

错误

MVSEP API 使用以下错误代码:

错误代码 含义
400 无效的 API 密钥

Quality Checker

GET QUEUE

To get information about Quality Checker queue entries, you need to make a GET call to the following url:
https://mvsep.com/api/quality_checker/queue

查询参数

字段 类型 说明
start Integer (可选)起始点偏移量,默认为 0(即从最新开始)
limit Integer (可选)返回的最大结果数量,默认为 10,最大为 20。

Curl示例代码:

curl --location --request GET 'https://mvsep.com/api/quality_checker/queue?&start=1&limit=2'

响应详情:

以数组形式,返回数据:

键 值
id Identification number of entry
dataset_type Dataset Type chosen when adding the entry / Leaderboard type
  • 0 - Synth
  • 1 - Multi
  • 2 - Piano
  • 3 - Lead/Back Vocals
  • 4 - Guitar
  • 5 - Medley Vox
  • 6 - Strings
  • 7 - Wind
  • 8 - DNR v3 Test
  • 9 - Super Resolution Checker for Music
  • 10 - Drums Separation (5 stems)
  • 11 - Male/Female vocals separation
date Created date of the entry
algo_name Name of the used separation type/algorithm
main_text Description of the used separation type/algorithm
proc Status of the processing
  • 0 - Not processed
  • 1 - Processed
proc_remote_date Timestamp of processing date
msg Returns "OK" if successful, otherwise shows error message
metrics Information about entry metrics
sdr_* (wildcard) SDR of the corresponding stems
dataset_name Name of the dataset

GET LEADERBOARD

To get information about Quality Checker leaderboards, you need to make a GET call to the following url:
https://mvsep.com/api/quality_checker/leaderboard

查询参数

字段 类型 说明
dataset_type String Dataset Type, defaults to 0 if not given
  • 0 - Synth
  • 1 - Multi
  • 2 - Piano
  • 3 - Lead/Back Vocals
  • 4 - Guitar
  • 5 - Medley Vox
  • 6 - Strings
  • 7 - Wind
  • 8 - DNR v3 Test
  • 9 - Super Resolution Checker for Music
  • 10 - Drums Separation (5 stems)
  • 11 - Male/Female vocals separation
start Integer (可选)起始点偏移量,默认为 0(即从最新开始)
limit Integer (可选)返回的最大结果数量,默认为 10,最大为 20。
algo_name_filter String Filters Separation Type/Algorithm names by given text
sort String Sort by given SDR of the stem. You can get list of sortable values for given dataset type from sortables key in response

Curl示例代码:

curl --location --request GET 'https://mvsep.com/api/quality_checker/leaderboard?&start=1&limit=2&algo_name_filter=MVSep'

响应详情:

键 值
data Contains array of Quality Checker entries. Refer to GET QUEUE response from Quality Checker Section for more information
dataset_type Used dataset type key
sortables Keys that can be used to sort used dataset type by given stem in sequential order.
sort Used sorting key from sortable.

CREATE ENTRY

To create Quality Checker entry, you need to make a POST call to the following url:
https://mvsep.com/api/quality_checker/add

查询参数

字段 类型 说明
api_token String (required) 您的 API Key
zipfile Binary (required) Zip file to be processed
algo_name String (required) Name of the algorithm
main_text String (required) Description of the algorithm
dataset_type String Dataset type. Defaults to 0.
  • 0 - Synth
  • 1 - Multi
  • 2 - Piano
  • 3 - Lead/Back Vocals
  • 4 - Guitar
  • 5 - Medley Vox
  • 6 - Strings
  • 7 - Wind
  • 8 - DNR v3 Test
  • 9 - Super Resolution Checker for Music
  • 10 - Drums Separation (5 stems)
  • 11 - Male/Female vocals separation
password String (required) Password to delete algorithm
ensemble integer 0 - algorithm is not ensemble. 1 - algorithm is ensemble.

Curl示例代码:

curl --location 'https://mvsep.com/api/quality_checker/add' --form 'zipfile=@"/path/to/zipfile"' --form 'algo_name="Test Algorithm Name"' --form 'main_text="Test Algorithm Description"' --form 'dataset_type="2"' --form 'ensemble="1"' --form 'password="test"'

响应详情:

键 值
success
  • false - when the entry creation failed
  • true - when the entry was created successfully
data Contains information about entry
data -> id ID of the entry.
data -> link Link to the entry.

错误

MVSEP API 使用以下错误代码:

错误代码 含义
400 表单无效
401 未知或无效的api_token。如果您使用了未知的API密钥,则会出现此错误。

GET ENTRY

To get information about Quality Checker queue entry by id, you need to make a GET call to the following url:
https://mvsep.com/api/quality_checker/entry

查询参数

字段 类型 说明
id Integer (required) ID of the entry

Curl示例代码:

curl --location --request GET 'https://mvsep.com/api/quality_checker/entry?id=7675'

响应详情:

Response is the same as GET QUEUE response from Quality Checker Section, returns only one entry.

DELETE ENTRY

To delete Quality Checker queue entry by id and password, you need to make a GET call to the following url:
https://mvsep.com/api/quality_checker/delete

查询参数

字段 类型 说明
id Integer (required) ID of the entry
password String (required) Used password for the entry

Curl示例代码:

curl --location 'https://mvsep.com/api/quality_checker/delete' --form 'id="7720"' --form 'password="test"'

响应详情:

键 值
success
  • false - when the entry deletion failed
  • true - when the entry was deleted successfully
data->message Shows whether entry was deleted
MVSEP Logo

turbo@mvsep.com

高级功能

质量检查工具

算法

完整 API 文档

公司

隐私政策

服务条款

退款政策

Cookie Notice

其他

帮助我们翻译!

帮助我们推广!