feat(tts): add pocket-tts backend (#8018)

* feat(pocket-tts): add new backend Signed-off-by: Ettore Di Giacinto <mudler@localai.io> * Add to the gallery Signed-off-by: Ettore Di Giacinto <mudler@localai.io> * fixups Signed-off-by: Ettore Di Giacinto <mudler@localai.io> * Update docs Signed-off-by: Ettore Di Giacinto <mudler@localai.io> --------- Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
2026-04-21 13:27:21 +00:00 · 2026-01-13 23:35:19 +01:00 · 2026-01-13 23:35:19 +01:00 · a6ff354c86
commit a6ff354c86
parent 3a2be4df48
25 changed files with 847 additions and 17 deletions
--- a/.github/workflows/backend.yml
+++ b/.github/workflows/backend.yml
@ -105,6 +105,19 @@ jobs:
            dockerfile: "./backend/Dockerfile.python"
            context: "./"
            ubuntu-version: '2404'
+          - build-type: 'cublas'
+            cuda-major-version: "12"
+            cuda-minor-version: "9"
+            platforms: 'linux/amd64'
+            tag-latest: 'auto'
+            tag-suffix: '-gpu-nvidia-cuda-12-pocket-tts'
+            runs-on: 'ubuntu-latest'
+            base-image: "ubuntu:24.04"
+            skip-drivers: 'false'
+            backend: "pocket-tts"
+            dockerfile: "./backend/Dockerfile.python"
+            context: "./"
+            ubuntu-version: '2404'
          - build-type: 'cublas'
            cuda-major-version: "12"
            cuda-minor-version: "0"
@ -340,6 +353,19 @@ jobs:
            dockerfile: "./backend/Dockerfile.python"
            context: "./"
            ubuntu-version: '2404'
+          - build-type: 'cublas'
+            cuda-major-version: "13"
+            cuda-minor-version: "0"
+            platforms: 'linux/amd64'
+            tag-latest: 'auto'
+            tag-suffix: '-gpu-nvidia-cuda-13-pocket-tts'
+            runs-on: 'ubuntu-latest'
+            base-image: "ubuntu:24.04"
+            skip-drivers: 'false'
+            backend: "pocket-tts"
+            dockerfile: "./backend/Dockerfile.python"
+            context: "./"
+            ubuntu-version: '2404'
          - build-type: 'cublas'
            cuda-major-version: "13"
            cuda-minor-version: "0"
@ -405,6 +431,19 @@ jobs:
            backend: "vibevoice"
            dockerfile: "./backend/Dockerfile.python"
            context: "./"
+          - build-type: 'l4t'
+            cuda-major-version: "13"
+            cuda-minor-version: "0"
+            platforms: 'linux/arm64'
+            tag-latest: 'auto'
+            tag-suffix: '-nvidia-l4t-cuda-13-arm64-pocket-tts'
+            runs-on: 'ubuntu-24.04-arm'
+            base-image: "ubuntu:24.04"
+            skip-drivers: 'false'
+            ubuntu-version: '2404'
+            backend: "pocket-tts"
+            dockerfile: "./backend/Dockerfile.python"
+            context: "./"
          - build-type: 'l4t'
            cuda-major-version: "13"
            cuda-minor-version: "0"
@ -641,6 +680,19 @@ jobs:
            dockerfile: "./backend/Dockerfile.python"
            context: "./"
            ubuntu-version: '2404'
+          - build-type: 'hipblas'
+            cuda-major-version: ""
+            cuda-minor-version: ""
+            platforms: 'linux/amd64'
+            tag-latest: 'auto'
+            tag-suffix: '-gpu-rocm-hipblas-pocket-tts'
+            runs-on: 'arc-runner-set'
+            base-image: "rocm/dev-ubuntu-24.04:6.4.4"
+            skip-drivers: 'false'
+            backend: "pocket-tts"
+            dockerfile: "./backend/Dockerfile.python"
+            context: "./"
+            ubuntu-version: '2404'
          - build-type: 'hipblas'
            cuda-major-version: ""
            cuda-minor-version: ""
@ -772,6 +824,19 @@ jobs:
            dockerfile: "./backend/Dockerfile.python"
            context: "./"
            ubuntu-version: '2204'
+          - build-type: 'l4t'
+            cuda-major-version: "12"
+            cuda-minor-version: "0"
+            platforms: 'linux/arm64'
+            tag-latest: 'auto'
+            tag-suffix: '-nvidia-l4t-pocket-tts'
+            runs-on: 'ubuntu-24.04-arm'
+            base-image: "nvcr.io/nvidia/l4t-jetpack:r36.4.0"
+            skip-drivers: 'true'
+            backend: "pocket-tts"
+            dockerfile: "./backend/Dockerfile.python"
+            context: "./"
+            ubuntu-version: '2204'
          - build-type: 'l4t'
            cuda-major-version: "12"
            cuda-minor-version: "0"
@ -825,6 +890,19 @@ jobs:
            dockerfile: "./backend/Dockerfile.python"
            context: "./"
            ubuntu-version: '2404'
+          - build-type: 'intel'
+            cuda-major-version: ""
+            cuda-minor-version: ""
+            platforms: 'linux/amd64'
+            tag-latest: 'auto'
+            tag-suffix: '-gpu-intel-pocket-tts'
+            runs-on: 'arc-runner-set'
+            base-image: "intel/oneapi-basekit:2025.3.0-0-devel-ubuntu24.04"
+            skip-drivers: 'false'
+            backend: "pocket-tts"
+            dockerfile: "./backend/Dockerfile.python"
+            context: "./"
+            ubuntu-version: '2404'
          - build-type: 'intel'
            cuda-major-version: ""
            cuda-minor-version: ""
@ -1278,6 +1356,19 @@ jobs:
            dockerfile: "./backend/Dockerfile.python"
            context: "./"
            ubuntu-version: '2404'
+          - build-type: ''
+            cuda-major-version: ""
+            cuda-minor-version: ""
+            platforms: 'linux/amd64,linux/arm64'
+            tag-latest: 'auto'
+            tag-suffix: '-cpu-pocket-tts'
+            runs-on: 'ubuntu-latest'
+            base-image: "ubuntu:24.04"
+            skip-drivers: 'false'
+            backend: "pocket-tts"
+            dockerfile: "./backend/Dockerfile.python"
+            context: "./"
+            ubuntu-version: '2404'
  backend-jobs-darwin:
    uses: ./.github/workflows/backend_build_darwin.yml
    strategy:
--- a/.github/workflows/test-extra.yml
+++ b/.github/workflows/test-extra.yml
@ -265,4 +265,23 @@ jobs:
      - name: Test moonshine
        run: |
          make --jobs=5 --output-sync=target -C backend/python/moonshine
-          make --jobs=5 --output-sync=target -C backend/python/moonshine test
+          make --jobs=5 --output-sync=target -C backend/python/moonshine test
+  tests-pocket-tts:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Clone
+        uses: actions/checkout@v6
+        with:
+          submodules: true
+      - name: Dependencies
+        run: |
+          sudo apt-get update
+          sudo apt-get install build-essential ffmpeg
+          sudo apt-get install -y ca-certificates cmake curl patch python3-pip
+          # Install UV
+          curl -LsSf https://astral.sh/uv/install.sh | sh
+          pip install --user --no-cache-dir grpcio-tools==1.64.1
+      - name: Test pocket-tts
+        run: |
+          make --jobs=5 --output-sync=target -C backend/python/pocket-tts
+          make --jobs=5 --output-sync=target -C backend/python/pocket-tts test
--- a/30
+++ b/30
@ -1,5 +1,5 @@
 # Disable parallel execution for backend builds
-.NOTPARALLEL: backends/diffusers backends/llama-cpp backends/piper backends/stablediffusion-ggml backends/whisper backends/faster-whisper backends/silero-vad backends/local-store backends/huggingface backends/rfdetr backends/kitten-tts backends/kokoro backends/chatterbox backends/llama-cpp-darwin backends/neutts build-darwin-python-backend build-darwin-go-backend backends/mlx backends/diffuser-darwin backends/mlx-vlm backends/mlx-audio backends/stablediffusion-ggml-darwin backends/vllm backends/moonshine
+.NOTPARALLEL: backends/diffusers backends/llama-cpp backends/piper backends/stablediffusion-ggml backends/whisper backends/faster-whisper backends/silero-vad backends/local-store backends/huggingface backends/rfdetr backends/kitten-tts backends/kokoro backends/chatterbox backends/llama-cpp-darwin backends/neutts build-darwin-python-backend build-darwin-go-backend backends/mlx backends/diffuser-darwin backends/mlx-vlm backends/mlx-audio backends/stablediffusion-ggml-darwin backends/vllm backends/moonshine backends/pocket-tts

 GOCMD=go
 GOTEST=$(GOCMD) test
@ -9,7 +9,7 @@ LAUNCHER_BINARY_NAME=local-ai-launcher

 CUDA_MAJOR_VERSION?=13
 CUDA_MINOR_VERSION?=0
-UBUNTU_VERSION?=2204
+UBUNTU_VERSION?=2404
 UBUNTU_CODENAME?=noble

 GORELEASER?=
@ -316,6 +316,7 @@ prepare-test-extra: protogen-python
 	$(MAKE) -C backend/python/vllm
 	$(MAKE) -C backend/python/vibevoice
 	$(MAKE) -C backend/python/moonshine
+	$(MAKE) -C backend/python/pocket-tts

 test-extra: prepare-test-extra
 	$(MAKE) -C backend/python/transformers test
@ -324,6 +325,7 @@ test-extra: prepare-test-extra
 	$(MAKE) -C backend/python/vllm test
 	$(MAKE) -C backend/python/vibevoice test
 	$(MAKE) -C backend/python/moonshine test
+	$(MAKE) -C backend/python/pocket-tts test

 DOCKER_IMAGE?=local-ai
 DOCKER_AIO_IMAGE?=local-ai-aio
@ -447,17 +449,16 @@ BACKEND_FASTER_WHISPER = faster-whisper|python|.|false|true
 BACKEND_COQUI = coqui|python|.|false|true
 BACKEND_BARK = bark|python|.|false|true
 BACKEND_EXLLAMA2 = exllama2|python|.|false|true
-
-# Python backends with ./backend context
-BACKEND_RFDETR = rfdetr|python|./backend|false|true
-BACKEND_KITTEN_TTS = kitten-tts|python|./backend|false|true
-BACKEND_NEUTTS = neutts|python|./backend|false|true
-BACKEND_KOKORO = kokoro|python|./backend|false|true
-BACKEND_VLLM = vllm|python|./backend|false|true
-BACKEND_DIFFUSERS = diffusers|python|./backend|--progress=plain|true
-BACKEND_CHATTERBOX = chatterbox|python|./backend|false|true
-BACKEND_VIBEVOICE = vibevoice|python|./backend|--progress=plain|true
-BACKEND_MOONSHINE = moonshine|python|./backend|false|true
+BACKEND_RFDETR = rfdetr|python|.|false|true
+BACKEND_KITTEN_TTS = kitten-tts|python|.|false|true
+BACKEND_NEUTTS = neutts|python|.|false|true
+BACKEND_KOKORO = kokoro|python|.|false|true
+BACKEND_VLLM = vllm|python|.|false|true
+BACKEND_DIFFUSERS = diffusers|python|.|--progress=plain|true
+BACKEND_CHATTERBOX = chatterbox|python|.|false|true
+BACKEND_VIBEVOICE = vibevoice|python|.|--progress=plain|true
+BACKEND_MOONSHINE = moonshine|python|.|false|true
+BACKEND_POCKET_TTS = pocket-tts|python|.|false|true

 # Helper function to build docker image for a backend
 # Usage: $(call docker-build-backend,BACKEND_NAME,DOCKERFILE_TYPE,BUILD_CONTEXT,PROGRESS_FLAG,NEEDS_BACKEND_ARG)
@ -503,12 +504,13 @@ $(eval $(call generate-docker-build-target,$(BACKEND_DIFFUSERS)))
 $(eval $(call generate-docker-build-target,$(BACKEND_CHATTERBOX)))
 $(eval $(call generate-docker-build-target,$(BACKEND_VIBEVOICE)))
 $(eval $(call generate-docker-build-target,$(BACKEND_MOONSHINE)))
+$(eval $(call generate-docker-build-target,$(BACKEND_POCKET_TTS)))

 # Pattern rule for docker-save targets
 docker-save-%: backend-images
 	docker save local-ai-backend:$* -o backend-images/$*.tar

-docker-build-backends: docker-build-llama-cpp docker-build-rerankers docker-build-vllm docker-build-transformers docker-build-diffusers docker-build-kokoro docker-build-faster-whisper docker-build-coqui docker-build-bark docker-build-chatterbox docker-build-vibevoice docker-build-exllama2 docker-build-moonshine
+docker-build-backends: docker-build-llama-cpp docker-build-rerankers docker-build-vllm docker-build-transformers docker-build-diffusers docker-build-kokoro docker-build-faster-whisper docker-build-coqui docker-build-bark docker-build-chatterbox docker-build-vibevoice docker-build-exllama2 docker-build-moonshine docker-build-pocket-tts

 ########################################################
 ### END Backends
--- a/README.md
+++ b/README.md
@ -295,6 +295,7 @@ LocalAI supports a comprehensive range of AI backends with multiple acceleration
 | **silero-vad** | Voice Activity Detection | CPU |
 | **neutts** | Text-to-speech with voice cloning | CUDA 12/13, ROCm, CPU |
 | **vibevoice** | Real-time TTS with voice cloning | CUDA 12/13, ROCm, Intel, CPU |
+| **pocket-tts** | Lightweight CPU-based TTS | CUDA 12/13, ROCm, Intel, CPU |

 ### Image & Video Generation
 | Backend | Description | Acceleration Support |
@ -316,8 +317,8 @@ LocalAI supports a comprehensive range of AI backends with multiple acceleration
 |-------------------|-------------------|------------------|
 | **NVIDIA CUDA 12** | All CUDA-compatible backends | Nvidia hardware |
 | **NVIDIA CUDA 13** | All CUDA-compatible backends | Nvidia hardware |
-| **AMD ROCm** | llama.cpp, whisper, vllm, transformers, diffusers, rerankers, coqui, kokoro, bark, neutts, vibevoice | AMD Graphics |
-| **Intel oneAPI** | llama.cpp, whisper, stablediffusion, vllm, transformers, diffusers, rfdetr, rerankers, exllama2, coqui, kokoro, bark, vibevoice | Intel Arc, Intel iGPUs |
+| **AMD ROCm** | llama.cpp, whisper, vllm, transformers, diffusers, rerankers, coqui, kokoro, bark, neutts, vibevoice, pocket-tts | AMD Graphics |
+| **Intel oneAPI** | llama.cpp, whisper, stablediffusion, vllm, transformers, diffusers, rfdetr, rerankers, exllama2, coqui, kokoro, bark, vibevoice, pocket-tts | Intel Arc, Intel iGPUs |
 | **Apple Metal** | llama.cpp, whisper, diffusers, MLX, MLX-VLM, bark-cpp | Apple M1/M2/M3+ |
 | **Vulkan** | llama.cpp, whisper, stablediffusion | Cross-platform GPUs |
 | **NVIDIA Jetson (CUDA 12)** | llama.cpp, whisper, stablediffusion, diffusers, rfdetr | ARM64 embedded AI (AGX Orin, etc.) |
--- a/backend/index.yaml
+++ b/backend/index.yaml
@ -428,6 +428,28 @@
    nvidia-l4t-cuda-12: "nvidia-l4t-vibevoice"
    nvidia-l4t-cuda-13: "cuda13-nvidia-l4t-arm64-vibevoice"
  icon: https://avatars.githubusercontent.com/u/6154722?s=200&v=4
+- &pocket-tts
+  urls:
+    - https://github.com/kyutai-labs/pocket-tts
+  description: |
+    Pocket TTS is a lightweight text-to-speech model designed to run efficiently on CPUs.
+  tags:
+    - text-to-speech
+    - TTS
+  license: mit
+  name: "pocket-tts"
+  alias: "pocket-tts"
+  capabilities:
+    nvidia: "cuda12-pocket-tts"
+    intel: "intel-pocket-tts"
+    amd: "rocm-pocket-tts"
+    nvidia-l4t: "nvidia-l4t-pocket-tts"
+    default: "cpu-pocket-tts"
+    nvidia-cuda-13: "cuda13-pocket-tts"
+    nvidia-cuda-12: "cuda12-pocket-tts"
+    nvidia-l4t-cuda-12: "nvidia-l4t-pocket-tts"
+    nvidia-l4t-cuda-13: "cuda13-nvidia-l4t-arm64-pocket-tts"
+  icon: https://avatars.githubusercontent.com/u/6154722?s=200&v=4
 - &piper
  name: "piper"
  uri: "quay.io/go-skynet/local-ai-backends:latest-piper"
@ -1605,3 +1627,86 @@
  uri: "quay.io/go-skynet/local-ai-backends:master-nvidia-l4t-cuda-13-arm64-vibevoice"
  mirrors:
    - localai/localai-backends:master-nvidia-l4t-cuda-13-arm64-vibevoice
+## pocket-tts
+- !!merge <<: *pocket-tts
+  name: "pocket-tts-development"
+  capabilities:
+    nvidia: "cuda12-pocket-tts-development"
+    intel: "intel-pocket-tts-development"
+    amd: "rocm-pocket-tts-development"
+    nvidia-l4t: "nvidia-l4t-pocket-tts-development"
+    default: "cpu-pocket-tts-development"
+    nvidia-cuda-13: "cuda13-pocket-tts-development"
+    nvidia-cuda-12: "cuda12-pocket-tts-development"
+    nvidia-l4t-cuda-12: "nvidia-l4t-pocket-tts-development"
+    nvidia-l4t-cuda-13: "cuda13-nvidia-l4t-arm64-pocket-tts-development"
+- !!merge <<: *pocket-tts
+  name: "cpu-pocket-tts"
+  uri: "quay.io/go-skynet/local-ai-backends:latest-cpu-pocket-tts"
+  mirrors:
+    - localai/localai-backends:latest-cpu-pocket-tts
+- !!merge <<: *pocket-tts
+  name: "cpu-pocket-tts-development"
+  uri: "quay.io/go-skynet/local-ai-backends:master-cpu-pocket-tts"
+  mirrors:
+    - localai/localai-backends:master-cpu-pocket-tts
+- !!merge <<: *pocket-tts
+  name: "cuda12-pocket-tts"
+  uri: "quay.io/go-skynet/local-ai-backends:latest-gpu-nvidia-cuda-12-pocket-tts"
+  mirrors:
+    - localai/localai-backends:latest-gpu-nvidia-cuda-12-pocket-tts
+- !!merge <<: *pocket-tts
+  name: "cuda12-pocket-tts-development"
+  uri: "quay.io/go-skynet/local-ai-backends:master-gpu-nvidia-cuda-12-pocket-tts"
+  mirrors:
+    - localai/localai-backends:master-gpu-nvidia-cuda-12-pocket-tts
+- !!merge <<: *pocket-tts
+  name: "cuda13-pocket-tts"
+  uri: "quay.io/go-skynet/local-ai-backends:latest-gpu-nvidia-cuda-13-pocket-tts"
+  mirrors:
+    - localai/localai-backends:latest-gpu-nvidia-cuda-13-pocket-tts
+- !!merge <<: *pocket-tts
+  name: "cuda13-pocket-tts-development"
+  uri: "quay.io/go-skynet/local-ai-backends:master-gpu-nvidia-cuda-13-pocket-tts"
+  mirrors:
+    - localai/localai-backends:master-gpu-nvidia-cuda-13-pocket-tts
+- !!merge <<: *pocket-tts
+  name: "intel-pocket-tts"
+  uri: "quay.io/go-skynet/local-ai-backends:latest-gpu-intel-pocket-tts"
+  mirrors:
+    - localai/localai-backends:latest-gpu-intel-pocket-tts
+- !!merge <<: *pocket-tts
+  name: "intel-pocket-tts-development"
+  uri: "quay.io/go-skynet/local-ai-backends:master-gpu-intel-pocket-tts"
+  mirrors:
+    - localai/localai-backends:master-gpu-intel-pocket-tts
+- !!merge <<: *pocket-tts
+  name: "rocm-pocket-tts"
+  uri: "quay.io/go-skynet/local-ai-backends:latest-gpu-rocm-hipblas-pocket-tts"
+  mirrors:
+    - localai/localai-backends:latest-gpu-rocm-hipblas-pocket-tts
+- !!merge <<: *pocket-tts
+  name: "rocm-pocket-tts-development"
+  uri: "quay.io/go-skynet/local-ai-backends:master-gpu-rocm-hipblas-pocket-tts"
+  mirrors:
+    - localai/localai-backends:master-gpu-rocm-hipblas-pocket-tts
+- !!merge <<: *pocket-tts
+  name: "nvidia-l4t-pocket-tts"
+  uri: "quay.io/go-skynet/local-ai-backends:latest-nvidia-l4t-pocket-tts"
+  mirrors:
+    - localai/localai-backends:latest-nvidia-l4t-pocket-tts
+- !!merge <<: *pocket-tts
+  name: "nvidia-l4t-pocket-tts-development"
+  uri: "quay.io/go-skynet/local-ai-backends:master-nvidia-l4t-pocket-tts"
+  mirrors:
+    - localai/localai-backends:master-nvidia-l4t-pocket-tts
+- !!merge <<: *pocket-tts
+  name: "cuda13-nvidia-l4t-arm64-pocket-tts"
+  uri: "quay.io/go-skynet/local-ai-backends:latest-nvidia-l4t-cuda-13-arm64-pocket-tts"
+  mirrors:
+    - localai/localai-backends:latest-nvidia-l4t-cuda-13-arm64-pocket-tts
+- !!merge <<: *pocket-tts
+  name: "cuda13-nvidia-l4t-arm64-pocket-tts-development"
+  uri: "quay.io/go-skynet/local-ai-backends:master-nvidia-l4t-cuda-13-arm64-pocket-tts"
+  mirrors:
+    - localai/localai-backends:master-nvidia-l4t-cuda-13-arm64-pocket-tts
--- a/backend/python/pocket-tts/Makefile
+++ b/backend/python/pocket-tts/Makefile
@ -0,0 +1,23 @@
+.PHONY: pocket-tts
+pocket-tts:
+	bash install.sh
+
+.PHONY: run
+run: pocket-tts
+	@echo "Running pocket-tts..."
+	bash run.sh
+	@echo "pocket-tts run."
+
+.PHONY: test
+test: pocket-tts
+	@echo "Testing pocket-tts..."
+	bash test.sh
+	@echo "pocket-tts tested."
+
+.PHONY: protogen-clean
+protogen-clean:
+	$(RM) backend_pb2_grpc.py backend_pb2.py
+
+.PHONY: clean
+clean: protogen-clean
+	rm -rf venv __pycache__
--- a/backend/python/pocket-tts/backend.py
+++ b/backend/python/pocket-tts/backend.py
@ -0,0 +1,255 @@
+#!/usr/bin/env python3
+"""
+This is an extra gRPC server of LocalAI for Pocket TTS
+"""
+from concurrent import futures
+import time
+import argparse
+import signal
+import sys
+import os
+import traceback
+import scipy.io.wavfile
+import backend_pb2
+import backend_pb2_grpc
+import torch
+from pocket_tts import TTSModel
+
+import grpc
+
+def is_float(s):
+    """Check if a string can be converted to float."""
+    try:
+        float(s)
+        return True
+    except ValueError:
+        return False
+
+def is_int(s):
+    """Check if a string can be converted to int."""
+    try:
+        int(s)
+        return True
+    except ValueError:
+        return False
+
+_ONE_DAY_IN_SECONDS = 60 * 60 * 24
+
+# If MAX_WORKERS are specified in the environment use it, otherwise default to 1
+MAX_WORKERS = int(os.environ.get('PYTHON_GRPC_MAX_WORKERS', '1'))
+
+# Implement the BackendServicer class with the service methods
+class BackendServicer(backend_pb2_grpc.BackendServicer):
+    """
+    BackendServicer is the class that implements the gRPC service
+    """
+    def Health(self, request, context):
+        return backend_pb2.Reply(message=bytes("OK", 'utf-8'))
+    
+    def LoadModel(self, request, context):
+        # Get device
+        if torch.cuda.is_available():
+            print("CUDA is available", file=sys.stderr)
+            device = "cuda"
+        else:
+            print("CUDA is not available", file=sys.stderr)
+            device = "cpu"
+        mps_available = hasattr(torch.backends, "mps") and torch.backends.mps.is_available()
+        if mps_available:
+            device = "mps"
+        if not torch.cuda.is_available() and request.CUDA:
+            return backend_pb2.Result(success=False, message="CUDA is not available")
+
+        # Normalize potential 'mpx' typo to 'mps'
+        if device == "mpx":
+            print("Note: device 'mpx' detected, treating it as 'mps'.", file=sys.stderr)
+            device = "mps"
+        
+        # Validate mps availability if requested
+        if device == "mps" and not torch.backends.mps.is_available():
+            print("Warning: MPS not available. Falling back to CPU.", file=sys.stderr)
+            device = "cpu"
+
+        self.device = device
+
+        options = request.Options
+
+        # empty dict
+        self.options = {}
+
+        # The options are a list of strings in this form optname:optvalue
+        # We are storing all the options in a dict so we can use it later when
+        # generating the audio
+        for opt in options:
+            if ":" not in opt:
+                continue
+            key, value = opt.split(":", 1)  # Split only on first colon
+            # if value is a number, convert it to the appropriate type
+            if is_float(value):
+                value = float(value)
+            elif is_int(value):
+                value = int(value)
+            elif value.lower() in ["true", "false"]:
+                value = value.lower() == "true"
+            self.options[key] = value
+
+        # Default voice for caching
+        self.default_voice_url = self.options.get("default_voice", None)
+        self._voice_cache = {}
+
+        try:
+            print("Loading Pocket TTS model", file=sys.stderr)
+            self.tts_model = TTSModel.load_model()
+            print(f"Model loaded successfully. Sample rate: {self.tts_model.sample_rate}", file=sys.stderr)
+
+            # Pre-load default voice if specified
+            if self.default_voice_url:
+                try:
+                    print(f"Pre-loading default voice: {self.default_voice_url}", file=sys.stderr)
+                    voice_state = self.tts_model.get_state_for_audio_prompt(self.default_voice_url)
+                    self._voice_cache[self.default_voice_url] = voice_state
+                    print("Default voice loaded successfully", file=sys.stderr)
+                except Exception as e:
+                    print(f"Warning: Failed to pre-load default voice: {e}", file=sys.stderr)
+
+        except Exception as err:
+            return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
+        
+        return backend_pb2.Result(message="Model loaded successfully", success=True)
+
+    def _get_voice_state(self, voice_input):
+        """
+        Get voice state from cache or load it.
+        voice_input can be:
+        - HuggingFace URL (e.g., hf://kyutai/tts-voices/alba-mackenna/casual.wav)
+        - Local file path
+        - None (use default)
+        """
+        # Use default if no voice specified
+        if not voice_input:
+            voice_input = self.default_voice_url
+
+        if not voice_input:
+            return None
+
+        # Check cache first
+        if voice_input in self._voice_cache:
+            return self._voice_cache[voice_input]
+
+        # Load voice state
+        try:
+            print(f"Loading voice from: {voice_input}", file=sys.stderr)
+            voice_state = self.tts_model.get_state_for_audio_prompt(voice_input)
+            self._voice_cache[voice_input] = voice_state
+            return voice_state
+        except Exception as e:
+            print(f"Error loading voice from {voice_input}: {e}", file=sys.stderr)
+            return None
+
+    def TTS(self, request, context):
+        try:
+            # Determine voice input
+            # Priority: request.voice > AudioPath (from ModelOptions) > default
+            voice_input = None
+            
+            if request.voice:
+                voice_input = request.voice
+            elif hasattr(request, 'AudioPath') and request.AudioPath:
+                # Use AudioPath as voice file
+                if os.path.isabs(request.AudioPath):
+                    voice_input = request.AudioPath
+                elif hasattr(request, 'ModelFile') and request.ModelFile:
+                    model_file_base = os.path.dirname(request.ModelFile)
+                    voice_input = os.path.join(model_file_base, request.AudioPath)
+                elif hasattr(request, 'ModelPath') and request.ModelPath:
+                    voice_input = os.path.join(request.ModelPath, request.AudioPath)
+                else:
+                    voice_input = request.AudioPath
+
+            # Get voice state
+            voice_state = self._get_voice_state(voice_input)
+            if voice_state is None:
+                return backend_pb2.Result(
+                    success=False,
+                    message=f"Voice not found or failed to load: {voice_input}. Please provide a valid voice URL or file path."
+                )
+
+            # Prepare text
+            text = request.text.strip()
+
+            if not text:
+                return backend_pb2.Result(
+                    success=False,
+                    message="Text is empty"
+                )
+
+            print(f"Generating audio for text: {text[:50]}...", file=sys.stderr)
+
+            # Generate audio
+            audio = self.tts_model.generate_audio(voice_state, text)
+
+            # Audio is a 1D torch tensor containing PCM data
+            if audio is None or audio.numel() == 0:
+                return backend_pb2.Result(
+                    success=False,
+                    message="No audio generated"
+                )
+
+            # Save audio to file
+            output_path = request.dst
+            if not output_path:
+                output_path = "/tmp/pocket-tts-output.wav"
+
+            # Ensure output directory exists
+            output_dir = os.path.dirname(output_path)
+            if output_dir and not os.path.exists(output_dir):
+                os.makedirs(output_dir, exist_ok=True)
+
+            # Convert torch tensor to numpy and save
+            audio_numpy = audio.numpy()
+            scipy.io.wavfile.write(output_path, self.tts_model.sample_rate, audio_numpy)
+            print(f"Saved audio to {output_path}", file=sys.stderr)
+
+        except Exception as err:
+            print(f"Error in TTS: {err}", file=sys.stderr)
+            print(traceback.format_exc(), file=sys.stderr)
+            return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
+        
+        return backend_pb2.Result(success=True)
+
+def serve(address):
+    server = grpc.server(futures.ThreadPoolExecutor(max_workers=MAX_WORKERS),
+        options=[
+            ('grpc.max_message_length', 50 * 1024 * 1024),  # 50MB
+            ('grpc.max_send_message_length', 50 * 1024 * 1024),  # 50MB
+            ('grpc.max_receive_message_length', 50 * 1024 * 1024),  # 50MB
+        ])
+    backend_pb2_grpc.add_BackendServicer_to_server(BackendServicer(), server)
+    server.add_insecure_port(address)
+    server.start()
+    print("Server started. Listening on: " + address, file=sys.stderr)
+
+    # Define the signal handler function
+    def signal_handler(sig, frame):
+        print("Received termination signal. Shutting down...")
+        server.stop(0)
+        sys.exit(0)
+
+    # Set the signal handlers for SIGINT and SIGTERM
+    signal.signal(signal.SIGINT, signal_handler)
+    signal.signal(signal.SIGTERM, signal_handler)
+
+    try:
+        while True:
+            time.sleep(_ONE_DAY_IN_SECONDS)
+    except KeyboardInterrupt:
+        server.stop(0)
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description="Run the gRPC server.")
+    parser.add_argument(
+        "--addr", default="localhost:50051", help="The address to bind the server to."
+    )
+    args = parser.parse_args()
+
+    serve(args.addr)
--- a/backend/python/pocket-tts/install.sh
+++ b/backend/python/pocket-tts/install.sh
@ -0,0 +1,30 @@
+#!/bin/bash
+set -e
+
+backend_dir=$(dirname $0)
+if [ -d $backend_dir/common ]; then
+    source $backend_dir/common/libbackend.sh
+else
+    source $backend_dir/../common/libbackend.sh
+fi
+
+# This is here because the Intel pip index is broken and returns 200 status codes for every package name, it just doesn't return any package links.
+# This makes uv think that the package exists in the Intel pip index, and by default it stops looking at other pip indexes once it finds a match.
+# We need uv to continue falling through to the pypi default index to find optimum[openvino] in the pypi index
+# the --upgrade actually allows us to *downgrade* torch to the version provided in the Intel pip index
+if [ "x${BUILD_PROFILE}" == "xintel" ]; then
+    EXTRA_PIP_INSTALL_FLAGS+=" --upgrade --index-strategy=unsafe-first-match"
+fi
+
+# Use python 3.12 for l4t
+if [ "x${BUILD_PROFILE}" == "xl4t13" ]; then
+  PYTHON_VERSION="3.12"
+  PYTHON_PATCH="12"
+  PY_STANDALONE_TAG="20251120"
+fi
+
+if [ "x${BUILD_PROFILE}" == "xl4t12" ]; then
+    USE_PIP=true
+fi
+
+installRequirements
--- a/backend/python/pocket-tts/protogen.sh
+++ b/backend/python/pocket-tts/protogen.sh
@ -0,0 +1,11 @@
+#!/bin/bash
+set -e
+
+backend_dir=$(dirname $0)
+if [ -d $backend_dir/common ]; then
+    source $backend_dir/common/libbackend.sh
+else
+    source $backend_dir/../common/libbackend.sh
+fi
+
+python3 -m grpc_tools.protoc -I../.. -I./ --python_out=. --grpc_python_out=. backend.proto
--- a/backend/python/pocket-tts/requirements-cpu.txt
+++ b/backend/python/pocket-tts/requirements-cpu.txt
@ -0,0 +1,4 @@
+--extra-index-url https://download.pytorch.org/whl/cpu
+pocket-tts
+scipy
+torch
--- a/backend/python/pocket-tts/requirements-cublas12.txt
+++ b/backend/python/pocket-tts/requirements-cublas12.txt
@ -0,0 +1,4 @@
+--extra-index-url https://download.pytorch.org/whl/cu121
+pocket-tts
+scipy
+torch
--- a/backend/python/pocket-tts/requirements-cublas13.txt
+++ b/backend/python/pocket-tts/requirements-cublas13.txt
@ -0,0 +1,4 @@
+--extra-index-url https://download.pytorch.org/whl/cu130
+pocket-tts
+scipy
+torch
--- a/backend/python/pocket-tts/requirements-hipblas.txt
+++ b/backend/python/pocket-tts/requirements-hipblas.txt
@ -0,0 +1,4 @@
+--extra-index-url https://download.pytorch.org/whl/rocm6.3
+pocket-tts
+scipy
+torch==2.7.1+rocm6.3
--- a/backend/python/pocket-tts/requirements-intel.txt
+++ b/backend/python/pocket-tts/requirements-intel.txt
@ -0,0 +1,4 @@
+--extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/
+pocket-tts
+scipy
+torch==2.5.1+cxx11.abi
--- a/backend/python/pocket-tts/requirements-l4t12.txt
+++ b/backend/python/pocket-tts/requirements-l4t12.txt
@ -0,0 +1,4 @@
+--extra-index-url https://pypi.jetson-ai-lab.io/jp6/cu129/
+pocket-tts
+scipy
+torch
--- a/backend/python/pocket-tts/requirements-l4t13.txt
+++ b/backend/python/pocket-tts/requirements-l4t13.txt
@ -0,0 +1,4 @@
+--extra-index-url https://download.pytorch.org/whl/cu130
+pocket-tts
+scipy
+torch
--- a/backend/python/pocket-tts/requirements-mps.txt
+++ b/backend/python/pocket-tts/requirements-mps.txt
@ -0,0 +1,4 @@
+pocket-tts
+scipy
+torch==2.7.1
+torchvision==0.22.1
--- a/backend/python/pocket-tts/requirements.txt
+++ b/backend/python/pocket-tts/requirements.txt
@ -0,0 +1,4 @@
+grpcio==1.71.0
+protobuf
+certifi
+packaging==24.1
--- a/backend/python/pocket-tts/run.sh
+++ b/backend/python/pocket-tts/run.sh
@ -0,0 +1,9 @@
+#!/bin/bash
+backend_dir=$(dirname $0)
+if [ -d $backend_dir/common ]; then
+    source $backend_dir/common/libbackend.sh
+else
+    source $backend_dir/../common/libbackend.sh
+fi
+
+startBackend $@
--- a/backend/python/pocket-tts/test.py
+++ b/backend/python/pocket-tts/test.py
@ -0,0 +1,141 @@
+"""
+A test script to test the gRPC service
+"""
+import unittest
+import subprocess
+import time
+import os
+import tempfile
+import backend_pb2
+import backend_pb2_grpc
+
+import grpc
+
+
+class TestBackendServicer(unittest.TestCase):
+    """
+    TestBackendServicer is the class that tests the gRPC service
+    """
+    def setUp(self):
+        """
+        This method sets up the gRPC service by starting the server
+        """
+        self.service = subprocess.Popen(["python3", "backend.py", "--addr", "localhost:50051"])
+        time.sleep(30)
+
+    def tearDown(self) -> None:
+        """
+        This method tears down the gRPC service by terminating the server
+        """
+        self.service.terminate()
+        self.service.wait()
+
+    def test_server_startup(self):
+        """
+        This method tests if the server starts up successfully
+        """
+        try:
+            self.setUp()
+            with grpc.insecure_channel("localhost:50051") as channel:
+                stub = backend_pb2_grpc.BackendStub(channel)
+                response = stub.Health(backend_pb2.HealthMessage())
+                self.assertEqual(response.message, b'OK')
+        except Exception as err:
+            print(err)
+            self.fail("Server failed to start")
+        finally:
+            self.tearDown()
+
+    def test_load_model(self):
+        """
+        This method tests if the model is loaded successfully
+        """
+        try:
+            self.setUp()
+            with grpc.insecure_channel("localhost:50051") as channel:
+                stub = backend_pb2_grpc.BackendStub(channel)
+                response = stub.LoadModel(backend_pb2.ModelOptions())
+                print(response)
+                self.assertTrue(response.success)
+                self.assertEqual(response.message, "Model loaded successfully")
+        except Exception as err:
+            print(err)
+            self.fail("LoadModel service failed")
+        finally:
+            self.tearDown()
+
+    def test_tts_with_hf_voice(self):
+        """
+        This method tests TTS generation with HuggingFace voice URL
+        """
+        try:
+            self.setUp()
+            with grpc.insecure_channel("localhost:50051") as channel:
+                stub = backend_pb2_grpc.BackendStub(channel)
+                # Load model
+                response = stub.LoadModel(backend_pb2.ModelOptions())
+                self.assertTrue(response.success)
+                
+                # Create temporary output file
+                with tempfile.NamedTemporaryFile(suffix='.wav', delete=False) as tmp_file:
+                    output_path = tmp_file.name
+                
+                # Test TTS with HuggingFace voice URL
+                tts_request = backend_pb2.TTSRequest(
+                    text="Hello world, this is a test.",
+                    dst=output_path,
+                    voice="azelma"
+                )
+                tts_response = stub.TTS(tts_request)
+                self.assertTrue(tts_response.success)
+                
+                # Verify output file exists and is not empty
+                self.assertTrue(os.path.exists(output_path))
+                self.assertGreater(os.path.getsize(output_path), 0)
+                
+                # Cleanup
+                os.unlink(output_path)
+        except Exception as err:
+            print(err)
+            self.fail("TTS service failed")
+        finally:
+            self.tearDown()
+
+    def test_tts_with_default_voice(self):
+        """
+        This method tests TTS generation with default voice (via AudioPath in LoadModel)
+        """
+        try:
+            self.setUp()
+            with grpc.insecure_channel("localhost:50051") as channel:
+                stub = backend_pb2_grpc.BackendStub(channel)
+                # Load model with default voice
+                load_request = backend_pb2.ModelOptions(
+                    Options=["default_voice:azelma"]
+                )
+                response = stub.LoadModel(load_request)
+                self.assertTrue(response.success)
+                
+                # Create temporary output file
+                with tempfile.NamedTemporaryFile(suffix='.wav', delete=False) as tmp_file:
+                    output_path = tmp_file.name
+                
+                # Test TTS without specifying voice (should use default)
+                tts_request = backend_pb2.TTSRequest(
+                    text="Hello world, this is a test.",
+                    dst=output_path
+                )
+                tts_response = stub.TTS(tts_request)
+                self.assertTrue(tts_response.success)
+                
+                # Verify output file exists and is not empty
+                self.assertTrue(os.path.exists(output_path))
+                self.assertGreater(os.path.getsize(output_path), 0)
+                
+                # Cleanup
+                os.unlink(output_path)
+        except Exception as err:
+            print(err)
+            self.fail("TTS service with default voice failed")
+        finally:
+            self.tearDown()
--- a/backend/python/pocket-tts/test.sh
+++ b/backend/python/pocket-tts/test.sh
@ -0,0 +1,11 @@
+#!/bin/bash
+set -e
+
+backend_dir=$(dirname $0)
+if [ -d $backend_dir/common ]; then
+    source $backend_dir/common/libbackend.sh
+else
+    source $backend_dir/../common/libbackend.sh
+fi
+
+runUnittests
--- a/docs/content/features/text-to-audio.md
+++ b/docs/content/features/text-to-audio.md
@ -164,6 +164,57 @@ curl http://localhost:8080/tts -H "Content-Type: application/json" -d '{
   }' | aplay
 ```

+### Pocket TTS
+
+[Pocket TTS](https://github.com/kyutai-labs/pocket-tts) is a lightweight text-to-speech model designed to run efficiently on CPUs. It supports voice cloning through HuggingFace voice URLs or local audio files.
+
+#### Setup
+
+Install the `pocket-tts` model in the Model gallery or run `local-ai run models install pocket-tts`.
+
+#### Usage
+
+Use the tts endpoint by specifying the pocket-tts backend:
+
+```
+curl http://localhost:8080/tts -H "Content-Type: application/json" -d '{         
+     "model": "pocket-tts",
+     "input":"Hello world, this is a test."
+   }' | aplay
+```
+
+#### Voice cloning
+
+Pocket TTS supports voice cloning through built-in voice names, HuggingFace URLs, or local audio files. You can configure a model with a specific voice:
+
+```yaml
+name: pocket-tts
+backend: pocket-tts
+tts:
+  voice: "azelma"  # Built-in voice name
+  # Or use HuggingFace URL: "hf://kyutai/tts-voices/alba-mackenna/casual.wav"
+  # Or use local file path: "path/to/voice.wav"
+  # Available built-in voices: alba, marius, javert, jean, fantine, cosette, eponine, azelma
+```
+
+You can also pre-load a default voice for faster first generation:
+
+```yaml
+name: pocket-tts
+backend: pocket-tts
+options:
+  - "default_voice:azelma"  # Pre-load this voice when model loads
+```
+
+Then you can use the model:
+
+```
+curl http://localhost:8080/tts -H "Content-Type: application/json" -d '{         
+     "model": "pocket-tts",
+     "input":"Hello world, this is a test."
+   }' | aplay
+```
+
 ### Vall-E-X

 [VALL-E-X](https://github.com/Plachtaa/VALL-E-X) is an open source implementation of Microsoft's VALL-E X zero-shot TTS model.
--- a/docs/content/reference/compatibility-table.md
+++ b/docs/content/reference/compatibility-table.md
@ -42,6 +42,7 @@ LocalAI will attempt to automatically load models which are not explicitly confi
 | [silero-vad](https://github.com/snakers4/silero-vad) with [Golang bindings](https://github.com/streamer45/silero-vad-go) | Silero VAD    | no                       | Voice Activity Detection    | no                               | no                   | CPU |
 | [neutts](https://github.com/neuphonic/neuttsair) | NeuTTSAir    | no                       | Text-to-speech with voice cloning    | no                               | no                   | CUDA 12/13, ROCm, CPU |
 | [vibevoice](https://github.com/microsoft/VibeVoice) | VibeVoice-Realtime    | no                       | Real-time text-to-speech with voice cloning    | no                               | no                   | CUDA 12/13, ROCm, Intel, CPU |
+| [pocket-tts](https://github.com/kyutai-labs/pocket-tts) | Pocket TTS    | no                       | Lightweight CPU-based text-to-speech with voice cloning    | no                               | no                   | CUDA 12/13, ROCm, Intel, CPU |
 | [mlx-audio](https://github.com/Blaizzy/mlx-audio) | MLX | no                       | Text-tospeech    | no                               | no                   | Metal (Apple Silicon) |

 ## Image & Video Generation
--- a/gallery/index.yaml
+++ b/gallery/index.yaml
@ -478,6 +478,16 @@
    - filename: voices/streaming_model/en-Davis_man.pt
      uri: https://raw.githubusercontent.com/microsoft/VibeVoice/main/demo/voices/streaming_model/en-Davis_man.pt
      sha256: 67561d63bfa2153616e4c02fd967007c182593fc53738a6ad94bf5f84e8832ac
+- &pocket-tts
+  url: "github:mudler/LocalAI/gallery/pocket-tts.yaml@master"
+  icon: https://avatars.githubusercontent.com/u/6154722?s=200&v=4
+  license: mit
+  tags:
+    - text-to-speech
+    - TTS
+  name: "pocket-tts"
+  urls:
+    - https://github.com/kyutai-labs/pocket-tts
 - &qwen3vl
  url: "github:mudler/LocalAI/gallery/qwen3.yaml@master"
  icon: https://cdn-avatars.huggingface.co/v1/production/uploads/620760a26e3b7210c2ff1943/-s1gyJfvbE1RgO5iBeNOi.png
--- a/gallery/pocket-tts.yaml
+++ b/gallery/pocket-tts.yaml
@ -0,0 +1,34 @@
+---
+name: localai
+
+config_file: |-
+    name: pocket-tts
+    backend: pocket-tts
+    description: |
+      Pocket TTS is a lightweight text-to-speech model designed to run efficiently on CPUs.
+      This model supports voice cloning through HuggingFace voice URLs or local audio files.
+
+    parameters:
+      model: ""
+
+    # TTS configuration
+    tts:
+      # Voice selection - can be:
+      # 1. Built-in voice name (e.g., "alba", "marius", "javert", "jean", "fantine", "cosette", "eponine", "azelma")
+      # 2. HuggingFace URL (e.g., "hf://kyutai/tts-voices/alba-mackenna/casual.wav")
+      # 3. Local file path (relative to model directory or absolute)
+      # voice: "azelma"
+      # Alternative: use audio_path to specify a voice file directly
+      # audio_path: "hf://kyutai/tts-voices/alba-mackenna/casual.wav"
+
+    known_usecases:
+      - tts
+
+    # Backend-specific options
+    # These are passed as "key:value" strings to the backend
+    options:
+      # Default voice to pre-load (optional)
+      # Can be a voice name or HuggingFace URL
+      # If set, this voice will be loaded when the model loads for faster first generation
+      - "default_voice:azelma"
+      # - "default_voice:hf://kyutai/tts-voices/alba-mackenna/casual.wav"