diff --git a/kustomize/bases/ollama/ollama-deployment.yaml b/kustomize/bases/ollama/ollama-deployment.yaml
index 417cb0d..8d6d6a3 100644
--- a/kustomize/bases/ollama/ollama-deployment.yaml
+++ b/kustomize/bases/ollama/ollama-deployment.yaml
@@ -31,7 +31,7 @@ spec:
           image: ollama/ollama
           env:
             - name: OLLAMA_KEEP_ALIVE
-              value: 36h
+              value: 12h
             # - name: HSA_OVERRIDE_GFX_VERSION
             #   value: 10.1.0
             # - name: HSA_ENABLE_SDMA
diff --git a/kustomize/bases/openedai-speech/kustomization.yaml b/kustomize/bases/openedai-speech/kustomization.yaml
new file mode 100644
index 0000000..44340f2
--- /dev/null
+++ b/kustomize/bases/openedai-speech/kustomization.yaml
@@ -0,0 +1,11 @@
+resources:
+  - openedai-speech-deployment.yaml
+
+namePrefix: openedai-speech-
+
+commonLabels:
+  app.kubernetes.io/name: openedai-speech
+
+# configMapGenerator:
+#   - name: server-config
+#     files: []
\ No newline at end of file
diff --git a/kustomize/bases/openedai-speech/openedai-speech-deployment.yaml b/kustomize/bases/openedai-speech/openedai-speech-deployment.yaml
new file mode 100644
index 0000000..b36ced9
--- /dev/null
+++ b/kustomize/bases/openedai-speech/openedai-speech-deployment.yaml
@@ -0,0 +1,93 @@
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: server
+  labels:
+      app.kubernetes.io/component: server
+spec:
+  replicas: 1
+  strategy:
+    type: Recreate
+  selector:
+    matchLabels:
+      app.kubernetes.io/component: server
+  template:
+    metadata:
+      labels:
+        app.kubernetes.io/component: server
+    spec:
+      affinity:
+        nodeAffinity:
+          requiredDuringSchedulingIgnoredDuringExecution:
+            nodeSelectorTerms:
+              - matchExpressions:
+                  # Image only supports amd64
+                  - key: kubernetes.io/arch
+                    operator: In
+                    values:
+                      - amd64
+      containers:
+        - name: openedai-speech
+          image: ghcr.io/matatonic/openedai-speech
+          ports:
+            - name: http
+              containerPort: 8000
+          resources:
+            requests:
+              cpu: 500m
+              memory: 1Gi
+            limits:
+              cpu: 1000m
+              memory: 1Gi
+          volumeMounts:
+            - name: server-voices-pv
+              mountPath: /app/voices
+            - name: server-config-pv
+              mountPath: /app/config
+      volumes:
+        - name: server-voices-pv
+          persistentVolumeClaim:
+            claimName: server-voices-pvc
+        - name: server-config-pv
+          persistentVolumeClaim:
+            claimName: server-config-pvc
+---
+apiVersion: v1
+kind: PersistentVolumeClaim
+metadata:
+  name: server-voices-pvc
+  labels:
+    app.kubernetes.io/name: server
+spec:
+  accessModes:
+    - ReadWriteOnce
+  resources:
+    requests:
+      storage: 5Gi
+---
+apiVersion: v1
+kind: PersistentVolumeClaim
+metadata:
+  name: server-config-pvc
+  labels:
+    app.kubernetes.io/name: server
+spec:
+  accessModes:
+    - ReadWriteOnce
+  resources:
+    requests:
+      storage: 1Gi
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: server
+  labels:
+      app.kubernetes.io/component: server
+spec:
+  selector:
+      app.kubernetes.io/component: server
+  ports:
+    - name: http
+      port: 8000
+      targetPort: http
\ No newline at end of file
diff --git a/kustomize/bases/openwebui/openwebui-deployment.yaml b/kustomize/bases/openwebui/openwebui-deployment.yaml
index 7821b23..24839af 100644
--- a/kustomize/bases/openwebui/openwebui-deployment.yaml
+++ b/kustomize/bases/openwebui/openwebui-deployment.yaml
@@ -25,10 +25,16 @@ spec:
                   fieldPath: metadata.namespace
             - name: OLLAMA_BASE_URL
               value: http://ollama-server.$(NAMESPACE).svc:11434
+            - name: AUDIO_TTS_OPENAI_API_BASE_URL
+              value: http://openedai-speech-server.$(NAMESPACE).svc:8000/v1
+            - name: AUDIO_TTS_OPENAI_API_KEY
+              value: ignored
             - name: RAG_EMBEDDING_ENGINE
               value: ollama
-            - name: AUDIO_STT_ENGINE
+            - name: AUDIO_TTS_ENGINE
               value: openai
+            # - name: AUDIO_STT_ENGINE
+            #   value: openai
             - name: ENABLE_SIGNUP
               value: "false" 
             - name: ENABLE_OAUTH_SIGNUP
@@ -57,8 +63,8 @@ spec:
             #   value: DEBUG
           resources:
             requests:
-              cpu: 250m
-              memory: 500Mi
+              cpu: 500m
+              memory: 1Gi
             limits:
               cpu: 500m
               memory: 1Gi
diff --git a/kustomize/overlays/llm/kustomization.yaml b/kustomize/overlays/llm/kustomization.yaml
index 699275c..bb8a58d 100644
--- a/kustomize/overlays/llm/kustomization.yaml
+++ b/kustomize/overlays/llm/kustomization.yaml
@@ -1,6 +1,7 @@
 resources:
   - namespace.yaml
   - ../../bases/ollama
+  - ../../bases/openedai-speech
   # - ../../bases/litellm
   - ../../bases/openwebui
   - ../../bases/sillytavern