Changed venv installation to use uv

.gitignore

@@ -103,7 +103,7 @@ celerybeat.pid
 
 # Environments
 .env
-.venv
+tools/venv
 env/
 venv/
 ENV/
@@ -127,7 +127,7 @@ dmypy.json
 
 # Pyre type checker
 .pyre/
-
+*.lock
 exp/
 downloads/
 data/

README.md

@@ -36,7 +36,7 @@ You can use the `_id` argument to specify which pre-trained model to use. If not
 > Since SHEET is a on-going project, if you use our pre-trained model in you paper, it is suggested to specify the version. For instance: `SHEET SSL-MOS v0.1.0`, `SHEET SSL-MOS v0.2.5`, etc.
 
 > [!TIP]
-> You don't need to install sheet following the [installation instructions](#instsallation). However, you might need to install the following:
+> You don't need to install sheet following the [installation instructions](#installation). However, you might need to install the following:
 > ```
 > sheet-sqa
 > huggingface_hub
@@ -63,18 +63,18 @@ You can use the `_id` argument to specify which pre-trained model to use. If not
 1.5806346
 ```
 
-## Instsallation 
+## installation 
 
 Full installation is needed if your goal is to do **training**.
 
 ### Editable installation with virtualenv 
 
-You don't need to prepare an environment (using conda, etc.) first. The following commands will automatically construct a virtual environment in `tools/`. When you run the recipes, the scripts will automatically activate the virtual environment.
+First install the uv package manager [here](https://docs.astral.sh/uv/getting-started/installation/). Then, use the following commands to automatically construct a virtual environment in `tools/venv`. When you run the recipes, the scripts will automatically activate the virtual environment.
 
 ```bash
 git clone https://github.com/unilight/sheet.git
-cd sheet/tools
-make
+cd sheet
+bash install.sh train
 ```
 
 ## Information

egs/bvcc/path.sh

@@ -6,7 +6,7 @@ if [ -e "${PRJ_ROOT}/tools/venv/bin/activate" ]; then
 fi
 
 MAIN_ROOT=$PWD/../..
-export PATH=$MAIN_ROOT/sheet/bin:$PATH
+export PATH=$MAIN_ROOT/src/sheet/bin:$PATH
 
 # python related
 export OMP_NUM_THREADS=1

hubconf.py

@@ -6,14 +6,19 @@
 """torch.hub configuration."""
 
 dependencies = ["yaml", "torch", "torchaudio", "sheet", "huggingface_hub"]
-
+import sys
 import os
 import torch
 import torch.nn.functional as F
 import torchaudio
 import yaml
 from huggingface_hub import hf_hub_download
 
+repo_root = os.path.dirname(os.path.abspath(__file__))
+src_path = os.path.join(repo_root, "src")
+if src_path not in sys.path:
+    sys.path.insert(0, 0, src_path)
+
 FS = 16000
 resamplers = {}
 MIN_REQUIRED_WAV_LENGTH = 1040

install.sh

@@ -0,0 +1,13 @@
+#!/bin/bash
+set -e
+
+uv venv tools/venv
+source tools/venv/bin/activate
+
+if [[ "$1" == "train" ]]; then
+    echo "Including 'train' extras..."
+    uv sync --extra train --active
+else
+    echo "Syncing without 'train' extras..."
+    uv sync --active
+fi

pyproject.toml

@@ -1,3 +1,43 @@
 [build-system]
-requires = ["setuptools"]
+requires = ["setuptools>=61", "wheel"]
 build-backend = "setuptools.build_meta"
+
+[project]
+name = "sheet"
+version = "0.2.5"
+description = "Speech Human Evaluation Estimation Toolkit (SHEET)"
+requires-python = "==3.10.13"
+
+authors = [
+    { name = "Wen-Chin Huang", email = "wen.chinhuang@g.sp.m.is.nagoya-u.ac.jp" }
+]
+
+dependencies = [
+    "torch==2.0.1",
+    "torchaudio==2.0.2",
+    "numpy==1.26.4",
+    "h5py>=3.15.1",
+    "pyyaml>=6.0.3",
+    "transformers>=4.57.1",
+    "scipy>=1.15.3",
+    "soundfile>=0.13.1",
+    "soxr>=1.0.0",
+    "wheel>=0.45.1",
+    "s3prl>=0.4.18",
+]
+
+[project.optional-dependencies]
+train = [
+    "matplotlib>=3.10.7",
+    "tqdm>=4.67.1",
+    "gdown",
+    "tensorboardx>=2.6.4",
+    "kaldiio>=2.14.1",
+    "humanfriendly>=10.0",
+    "prettytable>=3.16.0",
+]
+
+[tool.setuptools]
+packages = [
+    "sheet"
+]

sheet/models/utmos.py

@@ -43,6 +43,7 @@ def __init__(
         decoder_activation: str = "ReLU",
         output_type: str = "scalar",
         range_clipping: bool = True,
+        num_domains: int = None,
     ):
         super().__init__()  # this is needed! or else there will be an error.
         self.use_mean_listener = use_mean_listener