Merge pull request #1 from eliteprox/rename-audio-to-text

Rename pipeline speech-to-text to audio-to-text

runner/app/main.py

@@ -42,10 +42,10 @@ def load_pipeline(pipeline: str, model_id: str) -> any:
             from app.pipelines.image_to_video import ImageToVideoPipeline
 
             return ImageToVideoPipeline(model_id)
-        case "speech-to-text":
-            from app.pipelines.speech_to_text import SpeechToTextPipeline
+        case "audio-to-text":
+            from app.pipelines.audio_to_text import AudioToTextPipeline
 
-            return SpeechToTextPipeline(model_id)
+            return AudioToTextPipeline(model_id)
         case "frame-interpolation":
             raise NotImplementedError("frame-interpolation pipeline not implemented")
         case "upscale":
@@ -71,10 +71,10 @@ def load_route(pipeline: str) -> any:
             from app.routes import image_to_video
 
             return image_to_video.router
-        case "speech-to-text":
-            from app.routes import speech_to_text
+        case "audio-to-text":
+            from app.routes import audio_to_text
 
-            return speech_to_text.router
+            return audio_to_text.router
         case "frame-interpolation":
             raise NotImplementedError("frame-interpolation pipeline not implemented")
         case "upscale":

runner/app/pipelines/speech_to_text.py → runner/app/pipelines/audio_to_text.py

@@ -6,15 +6,14 @@
 
 from huggingface_hub import file_download
 import torch
-import PIL
 from typing import List
 import logging
 import os
 
 logger = logging.getLogger(__name__)
 
 
-class SpeechToTextPipeline(Pipeline):
+class AudioToTextPipeline(Pipeline):
     def __init__(self, model_id: str):
         # kwargs = {"cache_dir": get_model_dir()}
         kwargs = {}
@@ -25,21 +24,19 @@ def __init__(self, model_id: str):
         )
         folder_path = os.path.join(get_model_dir(), folder_name)
         # Load fp16 variant if fp16 safetensors files are found in cache
-        # Special case SDXL-Lightning because the safetensors files are fp16 but are not
-        # named properly right now
         has_fp16_variant = any(
             ".fp16.safetensors" in fname
             for _, _, files in os.walk(folder_path)
             for fname in files
         )
         if torch_device != "cpu" and has_fp16_variant:
-            logger.info("SpeechToTextPipeline loading fp16 variant for %s", model_id)
+            logger.info("AudioToTextPipeline loading fp16 variant for %s", model_id)
 
             kwargs["torch_dtype"] = torch.float16
             kwargs["variant"] = "fp16"
 
         if os.environ.get("BFLOAT16"):
-            logger.info("SpeechToTextPipeline using bfloat16 precision for %s", model_id)
+            logger.info("AudioToTextPipeline using bfloat16 precision for %s", model_id)
             kwargs["torch_dtype"] = torch.bfloat16
 
         self.model_id = model_id
@@ -81,4 +78,4 @@ def __call__(self, audio: str, **kwargs) -> List[File]:
         return result
 
     def __str__(self) -> str:
-        return f"SpeechToTextPipeline model_id={self.model_id}"
+        return f"AudioToTextPipeline model_id={self.model_id}"

runner/app/routes/speech_to_text.py → runner/app/routes/audio_to_text.py

@@ -17,9 +17,9 @@
 
 responses = {400: {"model": HTTPError}, 500: {"model": HTTPError}}
 
-@router.post("/speech-to-text", response_model=TextResponse, responses=responses)
-@router.post("/speech-to-text/", response_model=TextResponse, include_in_schema=False)
-async def speech_to_text(
+@router.post("/audio-to-text", response_model=TextResponse, responses=responses)
+@router.post("/audio-to-text/", response_model=TextResponse, include_in_schema=False)
+async def audio_to_text(
     audio: Annotated[UploadFile, File()],
     model_id: Annotated[str, Form()] = "",
     seed: Annotated[int, Form()] = None,

runner/dl_checkpoints.sh

@@ -30,7 +30,7 @@ function download_alpha_models() {
     # Download upscale models
     huggingface-cli download stabilityai/stable-diffusion-x4-upscaler --include "*.fp16.safetensors" --cache-dir models
 
-    # Download speech-to-text models.
+    # Download audio-to-text models.
     huggingface-cli download openai/whisper-large-v3 --include "*.safetensors" "*.json" --cache-dir models
 
     printf "\nDownloading token-gated models...\n"

runner/gen_openapi.py

@@ -5,9 +5,11 @@
 
 import yaml
 from app.main import app, use_route_names_as_operation_ids
-from app.routes import health, image_to_image, image_to_video, text_to_image, upscale, speech_to_text
+from app.routes import health, image_to_image, image_to_video, text_to_image, upscale
 from fastapi.openapi.utils import get_openapi
 
+from app.routes import audio_to_text
+
 # Specify Endpoints for OpenAPI schema generation.
 SERVERS = [
     {
@@ -77,7 +79,7 @@ def write_openapi(fname, entrypoint="runner"):
     app.include_router(image_to_image.router)
     app.include_router(image_to_video.router)
     app.include_router(upscale.router)
-    app.include_router(speech_to_text.router)
+    app.include_router(audio_to_text.router)
 
     use_route_names_as_operation_ids(app)
 

runner/openapi.json

@@ -282,15 +282,15 @@
                 ]
             }
         },
-        "/speech-to-text": {
+        "/audio-to-text": {
             "post": {
-                "summary": "Speech To Text",
-                "operationId": "speech_to_text",
+                "summary": "Audio To Text",
+                "operationId": "audio_to_text",
                 "requestBody": {
                     "content": {
                         "multipart/form-data": {
                             "schema": {
-                                "$ref": "#/components/schemas/Body_speech_to_text_speech_to_text_post"
+                                "$ref": "#/components/schemas/Body_audio_to_text_audio_to_text_post"
                             }
                         }
                     },
@@ -361,6 +361,29 @@
                 ],
                 "title": "APIError"
             },
+            "Body_audio_to_text_audio_to_text_post": {
+                "properties": {
+                    "audio": {
+                        "type": "string",
+                        "format": "binary",
+                        "title": "Audio"
+                    },
+                    "model_id": {
+                        "type": "string",
+                        "title": "Model Id",
+                        "default": ""
+                    },
+                    "seed": {
+                        "type": "integer",
+                        "title": "Seed"
+                    }
+                },
+                "type": "object",
+                "required": [
+                    "audio"
+                ],
+                "title": "Body_audio_to_text_audio_to_text_post"
+            },
             "Body_image_to_image_image_to_image_post": {
                 "properties": {
                     "prompt": {
@@ -472,29 +495,6 @@
                 ],
                 "title": "Body_image_to_video_image_to_video_post"
             },
-            "Body_speech_to_text_speech_to_text_post": {
-                "properties": {
-                    "audio": {
-                        "type": "string",
-                        "format": "binary",
-                        "title": "Audio"
-                    },
-                    "model_id": {
-                        "type": "string",
-                        "title": "Model Id",
-                        "default": ""
-                    },
-                    "seed": {
-                        "type": "integer",
-                        "title": "Seed"
-                    }
-                },
-                "type": "object",
-                "required": [
-                    "audio"
-                ],
-                "title": "Body_speech_to_text_speech_to_text_post"
-            },
             "Body_upscale_upscale_post": {
                 "properties": {
                     "prompt": {

worker/docker.go

@@ -34,7 +34,7 @@ var containerHostPorts = map[string]string{
 	"image-to-image": "8001",
 	"image-to-video": "8002",
 	"upscale":        "8003",
-	"speech-to-text": "8004",
+	"audio-to-text":  "8004",
 }
 
 type DockerManager struct {

worker/multipart.go

@@ -194,7 +194,7 @@ func NewUpscaleMultipartWriter(w io.Writer, req UpscaleMultipartRequestBody) (*m
 
 	return mw, nil
 }
-func NewSpeechToTextMultipartWriter(w io.Writer, req SpeechToTextMultipartRequestBody) (*multipart.Writer, error) {
+func NewAudioToTextMultipartWriter(w io.Writer, req AudioToTextMultipartRequestBody) (*multipart.Writer, error) {
 	mw := multipart.NewWriter(w)
 	writer, err := mw.CreateFormFile("audio", req.Audio.Filename())
 	if err != nil {