Input
Clean vocals, cover material, or separated vocal stems through an RVC v2 conversion chain
Audio formats
Output
Converted vocals, original vocal references, and downloadable audio for demos and listening tests
Use case
Voice-color conversion, cover demos, and character voice experiments
The AI cover pipeline combines separation, RVC voice conversion, and mixdown. The core inputs are a vocal file and target voice model.
RVC v2 converts the input vocal timbre into the selected character voice model.
RMVPE is the robust vocal pitch estimator used by the AI-RVC v1.5.1 default path.
MVSep 9205 supplies the lead/accompaniment-with-harmony leg, BS-RoFormer Leap XE 90 bands supplies vocals with harmony, and BS-RoFormer Leap Instrumental supplies the accompaniment reference.
The AI cover pipeline separates vocals/accompaniment before conversion and mixdown.
Scores apply only to the cited dataset, metric and model version. Leading this table does not mean a current world record or a result from this service. Capabilities and tests can justify prioritizing models without comparable scores.
The table includes only models with public weights or configurations that can run locally.
RMVPE paper rows for raw pitch accuracy, raw chroma accuracy, and overall accuracy on polyphonic singing datasets.
| Rank | Model | RPA (%) | Detailed metrics |
|---|---|---|---|
| MIR-1K #1 | RMVPEPolyTelkNetRMVPE Interspeech Table 1 — MIR-1K | 95.42Best | RCA (%)95.84BestOA (%)91.86Best |
| MIR-1K #2 | 91.05 | RCA (%)92.16OA (%)88.71 | |
| MIR-1K #3 | 88.63 | RCA (%)88.97OA (%)87.64 | |
| MIR-1K #4 | 82.28 | RCA (%)82.98OA (%)78.61 | |
| MIR-1K #5 | 77.29 | RCA (%)77.86OA (%)71.79 |
RMVPE paper rows for raw pitch accuracy, raw chroma accuracy, and overall accuracy on polyphonic singing datasets.
| Rank | Model | RPA (%) | Detailed metrics |
|---|---|---|---|
| MIR ST500 #1 | RMVPEPolyTelkNetRMVPE Interspeech Table 1 — MIR ST500 The paper states RMVPEPoly outperforms the second-best MIR ST500 model by 6.32% RPA. | 89.32Best | RCA (%)89.84BestOA (%)84.54Best |
| MIR ST500 #2 | 83.00 | RCA (%)83.35OA (%)81.80 | |
| MIR ST500 #3 | 81.61 | RCA (%)82.36OA (%)76.59 | |
| MIR ST500 #4 | 80.09 | RCA (%)80.38OA (%)82.86 | |
| MIR ST500 #5 | 63.85 | RCA (%)65.01OA (%)65.27 |
RMVPE paper rows for raw pitch accuracy, raw chroma accuracy, and overall accuracy on polyphonic singing datasets.
| Rank | Model | RPA (%) | Detailed metrics |
|---|---|---|---|
| Cmedia #1 | RMVPEPolyTelkNetRMVPE Interspeech Table 1 — Cmedia | 83.57Best | RCA (%)84.04BestOA (%)85.09Best |
| Cmedia #2 | 79.08 | RCA (%)79.64OA (%)81.83 | |
| Cmedia #3 | 79.00 | RCA (%)79.25OA (%)82.64 | |
| Cmedia #4 | 74.61 | RCA (%)75.05OA (%)75.19 | |
| Cmedia #5 | 59.69 | RCA (%)60.92OA (%)60.25 |
Leap XE 90 bands · Vocals SDR (dB)
Public sources checked: 2026-09-10
| Rank | Model | Vocals SDR (dB) | Detailed metrics |
|---|---|---|---|
| MVSep entry 10178 | BS-RoFormer Leap XE 90 bands (pcunwa)TelkNetMVSep 10178 metric | 11.7577Best | Instrumental SDR (dB)17.5303 |
| MVSep entry 10009 | BS PolarFormer public ONNX 62 bands (bgkb/ZFTurbo)MVSep 10009 metric | 11.7575 | Instrumental SDR (dB)18.0650Best |
| MVSep entry 10173 | BS-RoFormer Leap vocal-only (pcunwa)MVSep 10173 metric | 11.7398 | |
| MVSep entry 10146 | BS-RoFormer Leap 62 bands (pcunwa)MVSep 10146 metric | 11.7222 | Instrumental SDR (dB)17.7794 |
Public SDR is shown for Leap XE vocals. No matching public score is verified for the active Leap Instrumental checkpoint. Controlled-sample results are excluded from the public ranking.
Public sources checked: 2026-09-10
| Rank | Model | Instrumental SDR (dB) | Detailed metrics |
|---|---|---|---|
| MVSep entry 10009 | BS PolarFormer public ONNX 62 bands (bgkb/ZFTurbo)MVSep 10009 metric | 18.0650Best | Vocals SDR (dB)11.7575 |
| MVSep entry 10146 | BS-RoFormer Leap 62 bands (pcunwa)MVSep 10146 metric | 17.7794 | Vocals SDR (dB)11.7222 |
| MVSep entry 10178 | BS-RoFormer Leap XE 90 bands (pcunwa)MVSep 10178 metric | 17.5303 | Vocals SDR (dB)11.7577Best |
MVSep 9205 BS-Kar AVG · Lead vocals SDR (dB) / Back vocals SDR (dB)
Public sources checked: 2026-09-10
| Rank | Model | Lead vocals SDR (dB) | Detailed metrics |
|---|---|---|---|
| MVSep entry 9205 | BS-Kar-Gabox_IS + BS-Kar-Frazer&Becruily + BS-Kar-Anvuew (AVG)TelkNetMVSep 9205 metric | 10.4296Best | Back vocals + instrumental SDR (dB)15.8159Best |
| MVSep entry 9190 | BS-karaoke-Frazer&Becruily + BS-Karaoke-Anvuew (AVG)MVSep 9190 metric | 10.3596 | Back vocals + instrumental SDR (dB)15.7459 |
| MVSep entry 9209 | BS Roformer Karaoke by anvuewMVSep 9209 metric | 10.2267 | Back vocals + instrumental SDR (dB)15.6095 |
| MVSep entry 9212 | BS-Kar-Gabox_IS + BS-Kar-Frazer&Becruily + BS-Kar-Anvuew (Max Spec)MVSep 9212 metric | 10.1718 | Back vocals + instrumental SDR (dB)15.4680 |
| MVSep entry 9180 | BS-Roformer Karaoke anvuewMVSep 9180 metric | 10.1705 | Back vocals + instrumental SDR (dB)15.5568 |
| MVSep entry 9013 | BS-Roformer Karaoke (frazer & becruily)MVSep 9013 metric | 10.0640 | Back vocals + instrumental SDR (dB)15.4503 |
| MVSep entry 9198 | BS Roformer Karaoke GaboxMVSep 9198 metric | 9.8901 | Back vocals + instrumental SDR (dB)15.2763 |
IEEE/ACM TASLP - 2021
HuBERT provides self-supervised speech representations used by RVC-style voice conversion pipelines.
Interspeech - 2023
RMVPE is an Interspeech 2023 robust vocal pitch estimator designed for polyphonic music and used for quality-first F0 extraction.
Official repository - current
The RVC project is the official technical source for the VITS/HuBERT retrieval-based voice conversion workflow.
Working on a similar task? These guides list the matching tools, model boundaries, sources, and FAQs for each scenario.
Include the tool name, steps and the problem in your feedback. Screenshots can help us locate it.