Dataset Tool to add Timestamps (#121)

* add timestamp forced aligner

* download conda commands

* faster TimestampGenerationTask

Justfile

@@ -3,6 +3,7 @@ export WANDB_LOG_MODEL:="checkpoint"
 export PROJECT_DIR:="ultravox"
 export MCLOUD_CLUSTER:="r7z22p1"
 export MCLOUD_INSTANCE:="oci.bm.gpu.b4.8"
+export MFA_ENV_NAME:="aligner"
 
 default: format check test
 
@@ -62,3 +63,33 @@ run *FLAGS:
 
 mcloud *FLAGS:
     poetry run mcli interactive {{FLAGS}} --cluster ${MCLOUD_CLUSTER} --instance ${MCLOUD_INSTANCE}  --name `whoami` --command "bash -c \"$(cat setup.sh)\"" 
+
+@check_conda:
+    if ! command -v conda &> /dev/null; then  \
+        echo "Conda is not installed.";  \
+        mkdir -p ~/miniconda3;  \
+        if [ "$(uname)" = "Darwin" ]; then  \
+            echo "Downloading MacOS Miniconda.";  \
+            curl https://repo.anaconda.com/miniconda/Miniconda3-latest-MacOSX-arm64.sh -o ~/miniconda3/miniconda.sh;  \
+        elif [ "$(uname)" = "Linux" ]; then  \
+            echo "Downloading Linux Miniconda.";  \
+            wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda3/miniconda.sh  \
+        else  \
+            echo "Unknown operating system.";  \
+        fi;  \
+        bash ~/miniconda3/miniconda.sh -b -u -p ~/miniconda3;  \
+        rm ~/miniconda3/miniconda.sh;  \
+    else  \
+        echo "Conda is installed.";  \
+    fi
+
+@install_mfa: check_conda
+    if conda env list | grep -q "$MFA_ENV_NAME"; then  \
+        echo "Environment '$MFA_ENV_NAME' already exists.";  \
+    else  \
+        echo "Creating environment '$MFA_ENV_NAME'.";  \
+        conda create --name "$MFA_ENV_NAME" python=3.8 -y;  \
+        conda create -n "$MFA_ENV_NAME" -c conda-forge montreal-forced-aligner;  \
+        conda run -n "$MFA_ENV_NAME" mfa model download acoustic english_mfa;  \
+        conda run -n "$MFA_ENV_NAME" mfa model download dictionary english_mfa;  \
+    fi

pyproject.toml

@@ -30,6 +30,7 @@ wandb = "~0.17.1"
 sacrebleu = "^2.4.2"
 tenacity = "^9.0.0"
 evals = {git = "https://github.com/fixie-ai/evals", rev = "0c66bf85df7a4b903ecb202b23c2a826b749fd71"}
+praatio = "^6.2.0"
 
 [tool.poetry.group.dev.dependencies]
 black = "~24.4.2"