diff --git a/DOCKER_README.md b/DOCKER_README.md
index 496e7ae..098f883 100644
--- a/DOCKER_README.md
+++ b/DOCKER_README.md
@@ -54,6 +54,7 @@ following optional arguments:
 - `--rvc-voices` similar to voices, this option lets you pick the folder containing the [RVC models](https://github.com/erew123/alltalk_tts/wiki/RVC-(Retrieval%E2%80%90based-Voice-Conversion)).
     - Example: `docker-start.sh --rvc_voices /my/rvc/voices/dir`
 - `--no-ui` allows you to not expose port 7852 for the gradio interface and sets `launch_gradio` and `gradio_interface` to `false`.
+- `--auto-cleanup` periodically deletes all generated WAV files that are older than 1min.
 - `--with-multi-engine-manager` enables the use of the [multi engine manager (MEM)](https://github.com/erew123/alltalk_tts/wiki/Multi-Engine-Manager)
     which allows for more parallel requests. By default, one TTS engine is started. Optionally, you can pass the
     file path of a JSON file which can be a subset of `mem_config.json` with more fine-grained configuration options.
diff --git a/Dockerfile b/Dockerfile
index bcec9b5..2b0baf9 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -139,9 +139,19 @@ merge_json_files() {
 replace_env_vars docker_default_confignew.json
 merge_json_files confignew.json
 
+# Script for deleting WAV files that are older than 1 minute:
+cat << EOF2 > ${ALLTALK_DIR}/cleanup-wavs.sh
+#!/usr/bin/env bash
+[ \"\$ALLTALK_AUTO_CLEANUP\" = \"true\" ] && find ${ALLTALK_DIR}/outputs -type f -name "*.wav" -mmin +1 -exec rm {} \;
+EOF2
+chmod u+x ${ALLTALK_DIR}/cleanup-wavs.sh
+
+# Starting the cron job:
+cron
+
 source ${ALLTALK_DIR}/conda_env.sh
 
-if [ "\$ALLTALK_ENABLE_MULTI_ENGINE_MANAGER" = true ] ; then
+if [ "\$ALLTALK_ENABLE_MULTI_ENGINE_MANAGER" = "true" ] ; then
   echo "Starting alltalk using multi engine manager"
   replace_env_vars docker_default_mem_config.json
   merge_json_files mem_config.json
@@ -171,6 +181,25 @@ EOR
 
 COPY --chown=alltalk:alltalk . .
 
+RUN mkdir -p ${ALLTALK_DIR}/outputs ${ALLTALK_DIR}/.triton/autotune
+
+##############################################################################
+# Enable deepspeed for all models:
+##############################################################################
+RUN find . -name model_settings.json -exec sed -i -e 's/"deepspeed_enabled": false/"deepspeed_enabled": true/g' {} \;
+
+##############################################################################
+# Cronjob running every minute to delete output WAV files:
+##############################################################################
+USER root
+RUN <<EOR
+    echo "* * * * * ${ALLTALK_DIR}/cleanup-wavs.sh" > /etc/cron.d/cleanup-wavs
+    crontab -u alltalk /etc/cron.d/cleanup-wavs
+    chmod u+s /usr/sbin/cron
+EOR
+
+USER alltalk
+
 ##############################################################################
 # Create script to execute firstrun.py and run it:
 ##############################################################################
@@ -182,14 +211,7 @@ python ./system/config/firstrun.py $@' > ./start_firstrun.sh
 RUN chmod +x start_firstrun.sh
 RUN ./start_firstrun.sh --tts_model $TTS_MODEL
 
-RUN mkdir -p ${ALLTALK_DIR}/outputs ${ALLTALK_DIR}/.triton/autotune
-
-##############################################################################
-# Enable deepspeed for all models:
-##############################################################################
-RUN find . -name model_settings.json -exec sed -i -e 's/"deepspeed_enabled": false/"deepspeed_enabled": true/g' {} \;
-
 ##############################################################################
 # Start alltalk:
 ##############################################################################
-ENTRYPOINT ["sh", "-c", "./start_alltalk.sh"]
\ No newline at end of file
+ENTRYPOINT ["bash", "-c", "./start_alltalk.sh"]
\ No newline at end of file
diff --git a/docker-start.sh b/docker-start.sh
index 1eefaa6..e8b6bc3 100755
--- a/docker-start.sh
+++ b/docker-start.sh
@@ -9,6 +9,7 @@ MULTI_ENGINE_MANAGER_CONFIG=
 DOCKER_TAG=latest-xtts
 DOCKER_REPOSITORY=erew123
 ALLTALK_DIR=/home/alltalk
+ALLTALK_AUTO_CLEANUP=false
 declare -a ADDITIONAL_ARGS=()
 
 # Parse arguments
@@ -29,6 +30,9 @@ while [ "$#" -gt 0 ]; do
     --no-ui)
       ALLTALK_GRADIO_INTERFACE=false
       ;;
+    --auto-cleanup)
+      ALLTALK_AUTO_CLEANUP=true
+      ;;
     --with-multi-engine-manager)
       ALLTALK_ENABLE_MULTI_ENGINE_MANAGER=true
       if [ -n "$2" ] && ! [[ $2 =~ ^--.* ]]; then
@@ -91,6 +95,7 @@ fi
 DOCKER_ARGS+=( -e ALLTALK_ENABLE_MULTI_ENGINE_MANAGER=${ALLTALK_ENABLE_MULTI_ENGINE_MANAGER} )
 DOCKER_ARGS+=( -e ALLTALK_GRADIO_INTERFACE=${ALLTALK_GRADIO_INTERFACE} )
 DOCKER_ARGS+=( -e ALLTALK_LAUNCH_GRADIO=${ALLTALK_GRADIO_INTERFACE} )
+DOCKER_ARGS+=( -e ALLTALK_AUTO_CLEANUP=${ALLTALK_AUTO_CLEANUP} )
 
 
 docker run \
diff --git a/docker/base/Dockerfile b/docker/base/Dockerfile
index 46af3f3..91d119b 100644
--- a/docker/base/Dockerfile
+++ b/docker/base/Dockerfile
@@ -24,6 +24,7 @@ RUN <<EOR
     apt-get upgrade -y
     apt-get install --no-install-recommends -y \
       espeak-ng \
+      cron \
       curl \
       wget \
       jq \