更新多说话人训练webui，提供cli命令 / Update the multi-speaker training WebUI and provide CLI commands

2026-08-29 10:09:32 +02:00 · 2026-08-04 15:47:05 +08:00
parent ff8e396c3b
commit 81eed5e8f6
27 changed files with 6144 additions and 3784 deletions
--- a/docs/cn/cli.md
+++ b/docs/cn/cli.md
@@ -0,0 +1,362 @@
+# RVC 命令行训练与离线推理
+
+[English documentation](../en/cli.md)
+
+本文档只介绍 RVC 的训练和离线音色转换。PyMSS/UVR5 音源分离使用另一套 CLI。
+
+## 运行目录
+
+所有命令均在实际安装的项目根目录执行。将下面的占位内容替换为自己的路径：
+
+```powershell
+Set-Location "<RVC项目根目录>"
+```
+
+选择项目内置 Python 或已经安装好项目依赖的系统 Python。只保留其中一行：
+
+```powershell
+$PYTHON = "runtime\python.exe"  # 项目内置 Python
+# $PYTHON = "python"            # 系统 Python
+& $PYTHON --version
+```
+
+## 单说话人与多说话人的关系
+
+模型训练核心始终包含说话人 ID 嵌入，因此从模型结构角度看，多说话人训练覆盖单说话人场景。只含一个 ID 的多说话人清单也可以训练。
+
+当前项目仍然保留两套明确的数据约定：
+
+| 项目 | 单说话人 | 多说话人 |
+| --- | --- | --- |
+| 数据入口 | 目录第一层的音频文件 | 根目录下的 `名称_ID_重复次数` 子目录 |
+| 说话人 ID | 训练时统一指定一个 ID | 每条音频由 manifest 指定 ID |
+| 小模型元数据 | 不写 `speaker_info` | 写入名称和 ID 的 `speaker_info` |
+| 推理控件 | 默认 ID 0，可显式指定其他有效 ID | 显示名称和 ID，默认最小 ID |
+| 总特征文件 | `total_fea.npy` | 每人一份 `total_fea_spkid<ID>.npy` |
+| added 索引 | 普通索引文件名 | 每人一份 `_spkid<ID>.index` |
+
+只有一个音色时建议使用单说话人模式。需要在同一模型中按名称切换音色时使用多说话人模式。
+
+## 训练流程
+
+训练阶段依次为：数据切分、F0 提取、HuBERT 特征提取、模型训练、索引训练。
+
+### 单说话人数据切分
+
+```powershell
+& $PYTHON train\preprocess.py "训练集目录" 40000 8 "logs\实验名" False 3.7
+```
+
+参数依次为输入目录、采样率、进程数、实验目录、是否关闭多进程、切分参数。
+
+### 多说话人 manifest 与数据切分
+
+子目录格式为 `名称_ID_重复次数`，ID 范围为 `0~109`。
+
+```powershell
+& $PYTHON -c "from tools.multispeaker import build_manifest_from_root,write_manifest; write_manifest(r'logs\实验名',build_manifest_from_root(r'训练集总目录'))"
+
+& $PYTHON train\preprocess.py "" 40000 8 "logs\实验名" False 3.7 "logs\实验名\multispeaker_manifest.json"
+```
+
+重复次数只在最终 `filelist.txt` 中重复训练记录，不会重复执行切分、F0 或 HuBERT 特征提取。
+
+### F0 提取
+
+CPU + PM：
+
+```powershell
+& $PYTHON train\dataset\extract_f0.py cpu "logs\实验名" 8 pm
+```
+
+CPU + RMVPE：
+
+```powershell
+& $PYTHON train\dataset\extract_f0.py cpu "logs\实验名" 8 rmvpe
+```
+
+CUDA + RMVPE：
+
+```powershell
+& $PYTHON train\dataset\extract_f0.py cuda 1 0 0 "logs\实验名" true
+```
+
+CUDA 参数依次为总进程数、当前进程序号、GPU ID、实验目录和半精度开关。多个进程需要分别启动不同的当前进程序号。
+
+DirectML + RMVPE：
+
+```powershell
+& $PYTHON train\dataset\extract_f0.py dml "logs\实验名"
+```
+
+### HuBERT 特征提取
+
+CUDA：
+
+```powershell
+& $PYTHON train\dataset\extract_hubert_feature.py cuda:0 1 0 0 "logs\实验名" v2 true
+```
+
+CPU：
+
+```powershell
+& $PYTHON train\dataset\extract_hubert_feature.py cpu 1 0 "logs\实验名" v2 false
+```
+
+DirectML：
+
+```powershell
+& $PYTHON train\dataset\extract_hubert_feature.py privateuseone:0 1 0 "logs\实验名" v2 false
+```
+
+### 准备 `config.json` 和 `filelist.txt`
+
+`train.py` 启动前必须存在：
+
+```text
+logs\实验名\config.json
+logs\实验名\filelist.txt
+```
+
+单说话人带 F0 的每行格式：
+
+```text
+wav路径|HuBERT特征路径|粗F0路径|连续F0路径|说话人ID
+```
+
+多说话人带 F0 的每行格式：
+
+```text
+wav路径|HuBERT特征路径|粗F0路径|连续F0路径|说话人ID|说话人名称
+```
+
+无 F0 时分别去掉两个 F0 路径字段。多说话人的 `config.json` 还需要：
+
+```json
+{
+  "model": {
+    "spk_embed_dim": 110
+  },
+  "speaker_info": [
+    {"id": 0, "name": "说话人A"},
+    {"id": 1, "name": "说话人B"}
+  ]
+}
+```
+
+WebUI 的“训练模型”和“一键训练”会自动生成这两个文件；纯 CLI 流程需要在执行 `train.py` 前准备好它们。
+
+### 模型训练
+
+```powershell
+& $PYTHON train\train.py -e "实验名" -sr 40k -f0 1 -bs 8 -g 0 -te 200 -se 5 -pg assets\pretrained_v2\f0G40k.pth -pd assets\pretrained_v2\f0D40k.pth -l 0 -c 0 -sw 1 -v v2
+```
+
+常用参数：
+
+| 参数 | 含义 |
+| --- | --- |
+| `-e` | 实验名，对应 `logs\实验名` |
+| `-sr` | `32k`、`40k` 或 `48k` |
+| `-f0` | `1` 使用 F0，`0` 不使用 |
+| `-bs` | 每张卡的 batch size |
+| `-g` | GPU ID，例如 `0` 或 `0-1`；CPU 训练时可省略 |
+| `-te` | 总 epoch，最大 1200 |
+| `-se` | 保存间隔 epoch |
+| `-l` | `1` 只保留最新大 checkpoint |
+| `-c` | `1` 缓存训练集到显存 |
+| `-sw` | `1` 同时保存推理小模型 |
+| `-v` | `v1` 或 `v2` |
+
+### 索引训练
+
+单说话人：
+
+```powershell
+& $PYTHON train\train_index.py "实验名" v2 "assets\indices" 8 single
+```
+
+多说话人：
+
+```powershell
+& $PYTHON train\train_index.py "实验名" v2 "assets\indices" 8 multi
+```
+
+多说话人模式不是把所有人的特征混成一个索引。脚本会按说话人 ID 分组，为每个说话人分别生成 `total_fea_spkid<ID>.npy`、`trained_..._spkid<ID>.index` 和 `added_..._spkid<ID>.index`。因此模型中每个说话人都有自己的一份索引；推理选择说话人后，应使用该 ID 对应的 added 索引。
+
+自动按 manifest 判断：
+
+```powershell
+& $PYTHON train\train_index.py "实验名" v2 "assets\indices" 8 auto
+```
+
+## 离线推理 CLI
+
+入口为 `infer/cli.py`。
+
+查看帮助：
+
+```powershell
+& $PYTHON infer\cli.py --help
+```
+
+### 查看模型的说话人
+
+```powershell
+& $PYTHON infer\cli.py --model "assets\weights\模型.pth" --list-speakers
+```
+
+带 `speaker_info` 的多说话人模型会输出 ID 和名称。旧模型或单说话人模型会输出可用 ID 范围。
+
+### 单说话人模型推理
+
+```powershell
+& $PYTHON infer\cli.py `
+  --model "assets\weights\单说话人.pth" `
+  --input "D:\input.wav" `
+  --output "D:\output.wav" `
+  --pitch 0 `
+  --f0-method rmvpe `
+  --index-rate 0.75
+```
+
+单说话人默认使用 ID 0。旧模型含多个未命名 ID 时可通过 `--speaker-id` 指定。
+
+### 多说话人模型推理
+
+省略 `--speaker-id` 时使用 checkpoint 中最小的说话人 ID，并自动匹配该 ID 的 `_spkid<ID>.index`：
+
+```powershell
+& $PYTHON infer\cli.py `
+  --model "assets\weights\多说话人.pth" `
+  --input "D:\input.wav" `
+  --output "D:\output.wav" `
+  --index-rate 1
+```
+
+指定说话人：
+
+```powershell
+& $PYTHON infer\cli.py `
+  --model "assets\weights\多说话人.pth" `
+  --speaker-id 1 `
+  --input "D:\input.wav" `
+  --output "D:\output.wav" `
+  --index-rate 1
+```
+
+### 批量推理
+
+目录输入只扫描当前层；加入 `--recursive` 后递归扫描并保留相对目录：
+
+```powershell
+& $PYTHON infer\cli.py `
+  --model "assets\weights\模型.pth" `
+  --speaker-id 0 `
+  --input "D:\input_audio" `
+  --output "D:\output_audio" `
+  --format flac `
+  --recursive `
+  --overwrite
+```
+
+支持的输出格式为 `wav`、`flac`、`mp3` 和 `m4a`。输入扩展名判断不区分大小写。
+
+### 索引规则
+
+- 省略 `--index`：按模型名和说话人 ID 自动匹配 added 索引。
+- 指定 `--index`：使用该路径；trained 文件名会转换为对应 added 文件名。
+- `--index-rate` 大于 0 且索引不存在：命令以错误退出，不执行静默降级。
+- 不使用索引：传入 `--index-rate 0`。
+
+常用推理参数：
+
+| 参数 | 默认值 | 含义 |
+| --- | --- | --- |
+| `--speaker-id` | 单说话人 0；多说话人最小 ID | 推理说话人 |
+| `--pitch` | `0` | 半音变调 |
+| `--f0-method` | `rmvpe` | `pm` 或 `rmvpe` |
+| `--index-rate` | `0.75` | 索引检索占比 |
+| `--resample-sr` | `0` | 输出重采样率，0 表示模型采样率 |
+| `--rms-mix-rate` | `1.0` | 输出音量包络混合比例 |
+| `--protect` | `0.33` | 清辅音和呼吸声保护，范围 0~0.5 |
+| `--overwrite` | 关闭 | 覆盖已有输出 |
+
+正常完成返回退出码 0；发生推理失败返回 1；用户中断返回 130。
+
+## WebUI 当前 PyMSS 模型推理 CLI
+
+这里只列出 WebUI“人声伴奏分离&去混响”页面实际提供的 5 个模型。PyMSS 必须以模块方式调用：
+
+```powershell
+& $PYTHON -m tools.pymss.cli infer --help
+```
+
+WebUI 名称与 CLI 模型名对应如下：
+
+| WebUI 处理方式 | CLI 模型名 |
+| --- | --- |
+| 去混响 | `dereverb_mel_band_roformer_less_aggressive_anvuew_sdr_18.8050.ckpt` |
+| 去混响（激进） | `dereverb_mel_band_roformer_anvuew_sdr_19.1729.ckpt` |
+| 去伴奏 | `model_bs_roformer_ep_368_sdr_12.9628.ckpt` |
+| 去伴奏（激进） | `model_bs_roformer_ep_317_sdr_12.9755.ckpt` |
+| 提主旋律 | `model_mel_band_roformer_karaoke_aufr33_viperx_sdr_10.1956.ckpt` |
+
+`--input` 可以是单个音频文件或文件夹，`--output` 必须是输出文件夹。模型文件在 CLI 缓存中不存在时会自动下载。
+
+### 去混响
+
+```powershell
+& $PYTHON -m tools.pymss.cli infer `
+  "dereverb_mel_band_roformer_less_aggressive_anvuew_sdr_18.8050.ckpt" `
+  --input "<输入音频或目录>" `
+  --output "<输出目录>" `
+  --device auto `
+  --format flac
+```
+
+### 去混响（激进）
+
+```powershell
+& $PYTHON -m tools.pymss.cli infer `
+  "dereverb_mel_band_roformer_anvuew_sdr_19.1729.ckpt" `
+  --input "<输入音频或目录>" `
+  --output "<输出目录>" `
+  --device auto `
+  --format flac
+```
+
+### 去伴奏
+
+```powershell
+& $PYTHON -m tools.pymss.cli infer `
+  "model_bs_roformer_ep_368_sdr_12.9628.ckpt" `
+  --input "<输入音频或目录>" `
+  --output "<输出目录>" `
+  --device auto `
+  --format flac
+```
+
+### 去伴奏（激进）
+
+```powershell
+& $PYTHON -m tools.pymss.cli infer `
+  "model_bs_roformer_ep_317_sdr_12.9755.ckpt" `
+  --input "<输入音频或目录>" `
+  --output "<输出目录>" `
+  --device auto `
+  --format flac
+```
+
+### 提主旋律
+
+```powershell
+& $PYTHON -m tools.pymss.cli infer `
+  "model_mel_band_roformer_karaoke_aufr33_viperx_sdr_10.1956.ckpt" `
+  --input "<输入音频或目录>" `
+  --output "<输出目录>" `
+  --device auto `
+  --format flac
+```
+
+目录批处理时加入 `--save-as-folder`，可为每个输入音频建立单独的结果子目录。CUDA 可以将 `--device auto` 换成 `--device cuda --device-id 0`；CPU 可以换成 `--device cpu`。输出格式支持 `wav`、`flac`、`mp3` 和 `m4a`。
--- a/docs/en/cli.md
+++ b/docs/en/cli.md
@@ -0,0 +1,362 @@
+# RVC Command-Line Training and Offline Inference
+
+[中文文档](../cn/cli.md)
+
+This document covers RVC training and offline voice conversion. PyMSS/UVR5 source separation uses a separate CLI.
+
+## Working directory
+
+Run every command from the project's actual installation root. Replace the placeholder with your own path:
+
+```powershell
+Set-Location "<RVC_PROJECT_ROOT>"
+```
+
+Select either the bundled Python runtime or a system Python that has the project dependencies installed. Keep one assignment:
+
+```powershell
+$PYTHON = "runtime\python.exe"  # Bundled Python
+# $PYTHON = "python"            # System Python
+& $PYTHON --version
+```
+
+## Single-speaker and multi-speaker relationship
+
+The training architecture always includes a speaker-ID embedding. At the model level, multi-speaker training therefore includes the single-speaker case, and a multi-speaker manifest containing only one ID is valid.
+
+The project still maintains two explicit data conventions:
+
+| Area | Single-speaker | Multi-speaker |
+| --- | --- | --- |
+| Dataset input | Audio files directly inside one directory | `Name_ID_Repeat` subdirectories below a root directory |
+| Speaker ID | One ID assigned to every training record | One ID per manifest entry |
+| Small-model metadata | No `speaker_info` key | Named IDs stored in `speaker_info` |
+| Inference selection | ID 0 by default; another valid ID may be explicit | Names and IDs are exposed; the smallest ID is the default |
+| Total feature file | `total_fea.npy` | One `total_fea_spkid<ID>.npy` per speaker |
+| Added index | Standard index filename | One `_spkid<ID>.index` per speaker |
+
+Use single-speaker mode for one voice. Use multi-speaker mode when several named voices must be selectable from one model.
+
+## Training workflow
+
+The stages are dataset preprocessing, F0 extraction, HuBERT feature extraction, model training, and index training.
+
+### Single-speaker preprocessing
+
+```powershell
+& $PYTHON train\preprocess.py "DATASET_DIRECTORY" 40000 8 "logs\EXPERIMENT" False 3.7
+```
+
+The arguments are input directory, sample rate, worker count, experiment directory, disable-multiprocessing flag, and slicing parameter.
+
+### Multi-speaker manifest and preprocessing
+
+Subdirectory names use `Name_ID_Repeat`, and IDs range from `0` through `109`.
+
+```powershell
+& $PYTHON -c "from tools.multispeaker import build_manifest_from_root,write_manifest; write_manifest(r'logs\EXPERIMENT',build_manifest_from_root(r'DATASET_ROOT'))"
+
+& $PYTHON train\preprocess.py "" 40000 8 "logs\EXPERIMENT" False 3.7 "logs\EXPERIMENT\multispeaker_manifest.json"
+```
+
+The repeat count duplicates final `filelist.txt` records only. Preprocessing, F0 extraction, and HuBERT extraction are not repeated.
+
+### F0 extraction
+
+CPU with PM:
+
+```powershell
+& $PYTHON train\dataset\extract_f0.py cpu "logs\EXPERIMENT" 8 pm
+```
+
+CPU with RMVPE:
+
+```powershell
+& $PYTHON train\dataset\extract_f0.py cpu "logs\EXPERIMENT" 8 rmvpe
+```
+
+CUDA with RMVPE:
+
+```powershell
+& $PYTHON train\dataset\extract_f0.py cuda 1 0 0 "logs\EXPERIMENT" true
+```
+
+The CUDA arguments are total process count, current process index, GPU ID, experiment directory, and half-precision flag. Start one command for each process index when using several processes.
+
+DirectML with RMVPE:
+
+```powershell
+& $PYTHON train\dataset\extract_f0.py dml "logs\EXPERIMENT"
+```
+
+### HuBERT feature extraction
+
+CUDA:
+
+```powershell
+& $PYTHON train\dataset\extract_hubert_feature.py cuda:0 1 0 0 "logs\EXPERIMENT" v2 true
+```
+
+CPU:
+
+```powershell
+& $PYTHON train\dataset\extract_hubert_feature.py cpu 1 0 "logs\EXPERIMENT" v2 false
+```
+
+DirectML:
+
+```powershell
+& $PYTHON train\dataset\extract_hubert_feature.py privateuseone:0 1 0 "logs\EXPERIMENT" v2 false
+```
+
+### Prepare `config.json` and `filelist.txt`
+
+Before starting `train.py`, these files must exist:
+
+```text
+logs\EXPERIMENT\config.json
+logs\EXPERIMENT\filelist.txt
+```
+
+A single-speaker F0 record has this format:
+
+```text
+wav_path|HuBERT_feature_path|coarse_F0_path|continuous_F0_path|speaker_ID
+```
+
+A multi-speaker F0 record adds the speaker name:
+
+```text
+wav_path|HuBERT_feature_path|coarse_F0_path|continuous_F0_path|speaker_ID|speaker_name
+```
+
+Remove both F0 path fields for a non-F0 model. A multi-speaker `config.json` also needs:
+
+```json
+{
+  "model": {
+    "spk_embed_dim": 110
+  },
+  "speaker_info": [
+    {"id": 0, "name": "Speaker A"},
+    {"id": 1, "name": "Speaker B"}
+  ]
+}
+```
+
+The WebUI Model Training and One-click Training actions generate both files automatically. A pure CLI workflow prepares them before invoking `train.py`.
+
+### Model training
+
+```powershell
+& $PYTHON train\train.py -e "EXPERIMENT" -sr 40k -f0 1 -bs 8 -g 0 -te 200 -se 5 -pg assets\pretrained_v2\f0G40k.pth -pd assets\pretrained_v2\f0D40k.pth -l 0 -c 0 -sw 1 -v v2
+```
+
+Common options:
+
+| Option | Meaning |
+| --- | --- |
+| `-e` | Experiment name under `logs` |
+| `-sr` | `32k`, `40k`, or `48k` |
+| `-f0` | `1` enables F0; `0` disables it |
+| `-bs` | Batch size per GPU |
+| `-g` | GPU IDs such as `0` or `0-1`; omit for CPU training |
+| `-te` | Total epochs, up to 1200 |
+| `-se` | Checkpoint interval in epochs |
+| `-l` | `1` keeps only the latest large checkpoint |
+| `-c` | `1` caches the dataset in GPU memory |
+| `-sw` | `1` also saves inference-ready small models |
+| `-v` | `v1` or `v2` |
+
+### Index training
+
+Single-speaker:
+
+```powershell
+& $PYTHON train\train_index.py "EXPERIMENT" v2 "assets\indices" 8 single
+```
+
+Multi-speaker:
+
+```powershell
+& $PYTHON train\train_index.py "EXPERIMENT" v2 "assets\indices" 8 multi
+```
+
+Multi-speaker mode does not merge every speaker's features into one index. The script groups features by speaker ID and separately creates `total_fea_spkid<ID>.npy`, `trained_..._spkid<ID>.index`, and `added_..._spkid<ID>.index` for each speaker. Every speaker in the model therefore has its own index; after selecting a speaker for inference, use the added index for that ID.
+
+Automatic manifest detection:
+
+```powershell
+& $PYTHON train\train_index.py "EXPERIMENT" v2 "assets\indices" 8 auto
+```
+
+## Offline inference CLI
+
+The entry point is `infer/cli.py`.
+
+Show help:
+
+```powershell
+& $PYTHON infer\cli.py --help
+```
+
+### List model speakers
+
+```powershell
+& $PYTHON infer\cli.py --model "assets\weights\MODEL.pth" --list-speakers
+```
+
+Multi-speaker models with `speaker_info` print IDs and names. Older or single-speaker models print the valid ID range.
+
+### Single-speaker inference
+
+```powershell
+& $PYTHON infer\cli.py `
+  --model "assets\weights\SINGLE.pth" `
+  --input "D:\input.wav" `
+  --output "D:\output.wav" `
+  --pitch 0 `
+  --f0-method rmvpe `
+  --index-rate 0.75
+```
+
+Single-speaker models default to ID 0. Use `--speaker-id` for an older model that contains several unnamed IDs.
+
+### Multi-speaker inference
+
+When `--speaker-id` is omitted, the CLI selects the smallest declared ID and automatically matches its `_spkid<ID>.index`:
+
+```powershell
+& $PYTHON infer\cli.py `
+  --model "assets\weights\MULTI.pth" `
+  --input "D:\input.wav" `
+  --output "D:\output.wav" `
+  --index-rate 1
+```
+
+Select a specific speaker:
+
+```powershell
+& $PYTHON infer\cli.py `
+  --model "assets\weights\MULTI.pth" `
+  --speaker-id 1 `
+  --input "D:\input.wav" `
+  --output "D:\output.wav" `
+  --index-rate 1
+```
+
+### Batch inference
+
+Directory input scans direct child files. Add `--recursive` to scan subdirectories and preserve their relative layout:
+
+```powershell
+& $PYTHON infer\cli.py `
+  --model "assets\weights\MODEL.pth" `
+  --speaker-id 0 `
+  --input "D:\input_audio" `
+  --output "D:\output_audio" `
+  --format flac `
+  --recursive `
+  --overwrite
+```
+
+Output formats are `wav`, `flac`, `mp3`, and `m4a`. Input extension matching is case-insensitive.
+
+### Index behavior
+
+- Omit `--index` to match an added index by model name and speaker ID.
+- Supply `--index` to select a path explicitly. A trained filename is converted to its added counterpart.
+- If `--index-rate` is greater than zero and the index is missing, the command exits with an error instead of silently disabling retrieval.
+- Use `--index-rate 0` to disable retrieval.
+
+Common inference options:
+
+| Option | Default | Meaning |
+| --- | --- | --- |
+| `--speaker-id` | Single: 0; multi: smallest declared ID | Inference speaker |
+| `--pitch` | `0` | Pitch shift in semitones |
+| `--f0-method` | `rmvpe` | `pm` or `rmvpe` |
+| `--index-rate` | `0.75` | Retrieval blend ratio |
+| `--resample-sr` | `0` | Output sample rate; 0 keeps the model rate |
+| `--rms-mix-rate` | `1.0` | Output RMS-envelope blend ratio |
+| `--protect` | `0.33` | Consonant and breath protection, from 0 to 0.5 |
+| `--overwrite` | Off | Replace existing output files |
+
+Successful completion returns exit code 0, an inference failure returns 1, and user interruption returns 130.
+
+## CLI Inference For The Current WebUI PyMSS Models
+
+This section only covers the five models exposed by the WebUI's Vocal/Instrumental Separation and Dereverb page. Invoke PyMSS as a module:
+
+```powershell
+& $PYTHON -m tools.pymss.cli infer --help
+```
+
+The WebUI labels map to CLI catalog model names as follows:
+
+| WebUI action | CLI model name |
+| --- | --- |
+| Dereverb | `dereverb_mel_band_roformer_less_aggressive_anvuew_sdr_18.8050.ckpt` |
+| Dereverb (aggressive) | `dereverb_mel_band_roformer_anvuew_sdr_19.1729.ckpt` |
+| Remove instrumental | `model_bs_roformer_ep_368_sdr_12.9628.ckpt` |
+| Remove instrumental (aggressive) | `model_bs_roformer_ep_317_sdr_12.9755.ckpt` |
+| Extract lead vocal | `model_mel_band_roformer_karaoke_aufr33_viperx_sdr_10.1956.ckpt` |
+
+`--input` accepts one audio file or a directory, while `--output` is an output directory. Missing files in the CLI model cache are downloaded automatically.
+
+### Dereverb
+
+```powershell
+& $PYTHON -m tools.pymss.cli infer `
+  "dereverb_mel_band_roformer_less_aggressive_anvuew_sdr_18.8050.ckpt" `
+  --input "<input audio or directory>" `
+  --output "<output directory>" `
+  --device auto `
+  --format flac
+```
+
+### Dereverb (Aggressive)
+
+```powershell
+& $PYTHON -m tools.pymss.cli infer `
+  "dereverb_mel_band_roformer_anvuew_sdr_19.1729.ckpt" `
+  --input "<input audio or directory>" `
+  --output "<output directory>" `
+  --device auto `
+  --format flac
+```
+
+### Remove Instrumental
+
+```powershell
+& $PYTHON -m tools.pymss.cli infer `
+  "model_bs_roformer_ep_368_sdr_12.9628.ckpt" `
+  --input "<input audio or directory>" `
+  --output "<output directory>" `
+  --device auto `
+  --format flac
+```
+
+### Remove Instrumental (Aggressive)
+
+```powershell
+& $PYTHON -m tools.pymss.cli infer `
+  "model_bs_roformer_ep_317_sdr_12.9755.ckpt" `
+  --input "<input audio or directory>" `
+  --output "<output directory>" `
+  --device auto `
+  --format flac
+```
+
+### Extract Lead Vocal
+
+```powershell
+& $PYTHON -m tools.pymss.cli infer `
+  "model_mel_band_roformer_karaoke_aufr33_viperx_sdr_10.1956.ckpt" `
+  --input "<input audio or directory>" `
+  --output "<output directory>" `
+  --device auto `
+  --format flac
+```
+
+For directory batches, add `--save-as-folder` to create a separate result folder for each input. On CUDA, replace `--device auto` with `--device cuda --device-id 0`; for CPU, use `--device cpu`. Output formats are `wav`, `flac`, `mp3`, and `m4a`.