feat(chatterbox): support multilingual (#6240)

* feat(chatterbox): support multilingual

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* Add l4t support

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* Fixups

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

* fix: switch to fork

Until https://github.com/resemble-ai/chatterbox/pull/295 is merged

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

---------

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

.github/workflows/backend.yml

@@ -955,6 +955,18 @@ jobs:
             backend: "exllama2"
             dockerfile: "./backend/Dockerfile.python"
             context: "./backend"
+          - build-type: 'cublas'
+            cuda-major-version: "12"
+            cuda-minor-version: "0"
+            platforms: 'linux/arm64'
+            skip-drivers: 'true'
+            tag-latest: 'auto'
+            tag-suffix: '-nvidia-l4t-arm64-chatterbox'
+            base-image: "nvcr.io/nvidia/l4t-jetpack:r36.4.0"
+            runs-on: 'ubuntu-24.04-arm'
+            backend: "chatterbox"
+            dockerfile: "./backend/Dockerfile.python"
+            context: "./backend"
           # runs out of space on the runner
           # - build-type: 'hipblas'
           #   cuda-major-version: ""

Makefile

@@ -429,6 +429,9 @@ docker-build-kitten-tts:
 docker-save-kitten-tts: backend-images
 	docker save local-ai-backend:kitten-tts -o backend-images/kitten-tts.tar
 
+docker-save-chatterbox: backend-images
+	docker save local-ai-backend:chatterbox -o backend-images/chatterbox.tar
+
 docker-build-kokoro:
 	docker build --build-arg BUILD_TYPE=$(BUILD_TYPE) --build-arg BASE_IMAGE=$(BASE_IMAGE) -t local-ai-backend:kokoro -f backend/Dockerfile.python --build-arg BACKEND=kokoro ./backend
 

backend/index.yaml

@@ -353,6 +353,7 @@
     nvidia: "cuda12-chatterbox"
     metal: "metal-chatterbox"
     default: "cpu-chatterbox"
+    nvidia-l4t: "nvidia-l4t-arm64-chatterbox"
 - &piper
   name: "piper"
   uri: "quay.io/go-skynet/local-ai-backends:latest-piper"
@@ -1239,6 +1240,7 @@
     nvidia: "cuda12-chatterbox-development"
     metal: "metal-chatterbox-development"
     default: "cpu-chatterbox-development"
+    nvidia-l4t: "nvidia-l4t-arm64-chatterbox"
 - !!merge <<: *chatterbox
   name: "cpu-chatterbox"
   uri: "quay.io/go-skynet/local-ai-backends:latest-cpu-chatterbox"
@@ -1249,6 +1251,16 @@
   uri: "quay.io/go-skynet/local-ai-backends:master-cpu-chatterbox"
   mirrors:
     - localai/localai-backends:master-cpu-chatterbox
+- !!merge <<: *chatterbox
+  name: "nvidia-l4t-arm64-chatterbox"
+  uri: "quay.io/go-skynet/local-ai-backends:latest-gpu-nvidia-l4t-arm64-chatterbox"
+  mirrors:
+    - localai/localai-backends:latest-gpu-nvidia-l4t-arm64-chatterbox
+- !!merge <<: *chatterbox
+  name: "nvidia-l4t-arm64-chatterbox-development"
+  uri: "quay.io/go-skynet/local-ai-backends:master-gpu-nvidia-l4t-arm64-chatterbox"
+  mirrors:
+    - localai/localai-backends:master-gpu-nvidia-l4t-arm64-chatterbox
 - !!merge <<: *chatterbox
   name: "metal-chatterbox"
   uri: "quay.io/go-skynet/local-ai-backends:latest-metal-darwin-arm64-chatterbox"

backend/python/chatterbox/backend.py

@@ -14,9 +14,23 @@ import backend_pb2_grpc
 import torch
 import torchaudio as ta
 from chatterbox.tts import ChatterboxTTS
-
+from chatterbox.mtl_tts import ChatterboxMultilingualTTS
 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
 
@@ -47,6 +61,28 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
         if not torch.cuda.is_available() and request.CUDA:
             return backend_pb2.Result(success=False, message="CUDA is not available")
 
+
+        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 images
+        for opt in options:
+            if ":" not in opt:
+                continue
+            key, value = opt.split(":")
+            # 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
+
         self.AudioPath = None
 
         if os.path.isabs(request.AudioPath):
@@ -56,10 +92,14 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
             modelFileBase = os.path.dirname(request.ModelFile)
             # modify LoraAdapter to be relative to modelFileBase
             self.AudioPath = os.path.join(modelFileBase, request.AudioPath)
-
         try:
             print("Preparing models, please wait", file=sys.stderr)
-            self.model = ChatterboxTTS.from_pretrained(device=device)
+            if "multilingual" in self.options:
+                # remove key from options
+                del self.options["multilingual"]
+                self.model = ChatterboxMultilingualTTS.from_pretrained(device=device)
+            else:
+                self.model = ChatterboxTTS.from_pretrained(device=device)
         except Exception as err:
             return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
         # Implement your logic here for the LoadModel service
@@ -68,12 +108,18 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
 
     def TTS(self, request, context):
         try:
-            # Generate audio using ChatterboxTTS
+            kwargs = {}
+
+            if "language" in self.options:
+                kwargs["language_id"] = self.options["language"]
             if self.AudioPath is not None:
-                wav = self.model.generate(request.text, audio_prompt_path=self.AudioPath)
-            else:
-                wav = self.model.generate(request.text)
-            
+                kwargs["audio_prompt_path"] = self.AudioPath
+
+            # add options to kwargs
+            kwargs.update(self.options)
+
+            # Generate audio using ChatterboxTTS
+            wav = self.model.generate(request.text, **kwargs)
             # Save the generated audio
             ta.save(request.dst, wav, self.model.sr)
             

backend/python/chatterbox/install.sh

@@ -15,5 +15,6 @@ fi
 if [ "x${BUILD_PROFILE}" == "xintel" ]; then
     EXTRA_PIP_INSTALL_FLAGS+=" --upgrade --index-strategy=unsafe-first-match"
 fi
+EXTRA_PIP_INSTALL_FLAGS+=" --no-build-isolation"
 
 installRequirements

backend/python/chatterbox/requirements-cpu.txt

@@ -1,6 +1,8 @@
 --extra-index-url https://download.pytorch.org/whl/cpu
 accelerate
-torch==2.6.0
-torchaudio==2.6.0
-transformers==4.46.3
-chatterbox-tts==0.1.2
+torch
+torchaudio
+transformers
+# https://github.com/mudler/LocalAI/pull/6240#issuecomment-3329518289
+chatterbox-tts@git+https://git@github.com/mudler/chatterbox.git@faster
+#chatterbox-tts==0.1.4

backend/python/chatterbox/requirements-cublas11.txt

@@ -2,5 +2,6 @@
 torch==2.6.0+cu118
 torchaudio==2.6.0+cu118
 transformers==4.46.3
-chatterbox-tts==0.1.2
+# https://github.com/mudler/LocalAI/pull/6240#issuecomment-3329518289
+chatterbox-tts@git+https://git@github.com/mudler/chatterbox.git@faster
 accelerate

backend/python/chatterbox/requirements-cublas12.txt

@@ -1,5 +1,6 @@
-torch==2.6.0
-torchaudio==2.6.0
-transformers==4.46.3
-chatterbox-tts==0.1.2
+torch
+torchaudio
+transformers
+# https://github.com/mudler/LocalAI/pull/6240#issuecomment-3329518289
+chatterbox-tts@git+https://git@github.com/mudler/chatterbox.git@faster
 accelerate

backend/python/chatterbox/requirements-hipblas.txt

@@ -1,6 +1,7 @@
 --extra-index-url https://download.pytorch.org/whl/rocm6.0
 torch==2.6.0+rocm6.1
 torchaudio==2.6.0+rocm6.1
-transformers==4.46.3
-chatterbox-tts==0.1.2
+transformers
+# https://github.com/mudler/LocalAI/pull/6240#issuecomment-3329518289
+chatterbox-tts@git+https://git@github.com/mudler/chatterbox.git@faster
 accelerate

backend/python/chatterbox/requirements-intel.txt

@@ -2,8 +2,9 @@
 intel-extension-for-pytorch==2.3.110+xpu
 torch==2.3.1+cxx11.abi
 torchaudio==2.3.1+cxx11.abi
-transformers==4.46.3
-chatterbox-tts==0.1.2
+transformers
+# https://github.com/mudler/LocalAI/pull/6240#issuecomment-3329518289
+chatterbox-tts@git+https://git@github.com/mudler/chatterbox.git@faster
 accelerate
 oneccl_bind_pt==2.3.100+xpu
 optimum[openvino]

backend/python/chatterbox/requirements-l4t.txt

@@ -0,0 +1,6 @@
+--extra-index-url https://pypi.jetson-ai-lab.io/jp6/cu126/
+torch
+torchaudio
+transformers
+chatterbox-tts@git+https://git@github.com/mudler/chatterbox.git@faster
+accelerate