up

README.md

@@ -1,5 +1,5 @@
 ---
-title: Whisper French Demo
+title: Whisper German Demo
 emoji: 🤫
 colorFrom: indigo
 colorTo: red

run_demo.py

@@ -5,7 +5,7 @@ import pytube as pt
 from transformers import pipeline
 from huggingface_hub import model_info
 
-MODEL_NAME = "bofenghuang/whisper-medium-cv11-french-punct"
+MODEL_NAME = "bofenghuang/whisper-medium-cv11-german-punct"
 CHUNK_LENGTH_S = 30
 
 device = 0 if torch.cuda.is_available() else "cpu"
@@ -16,7 +16,7 @@ pipe = pipeline(
     device=device,
 )
 
-pipe.model.config.forced_decoder_ids = pipe.tokenizer.get_decoder_prompt_ids(language="fr", task="transcribe")
+pipe.model.config.forced_decoder_ids = pipe.tokenizer.get_decoder_prompt_ids(language="de", task="transcribe")
 
 def transcribe(microphone, file_upload):
     warn_output = ""

run_demo_multi_models.py

@@ -11,15 +11,12 @@ from transformers.utils.logging import disable_progress_bar
 warnings.filterwarnings("ignore")
 disable_progress_bar()
 
-DEFAULT_MODEL_NAME = "bofenghuang/whisper-medium-cv11-french-punct"
+DEFAULT_MODEL_NAME = "bofenghuang/whisper-medium-cv11-german-punct"
 MODEL_NAMES = [
     "openai/whisper-small",
     "openai/whisper-medium",
     "openai/whisper-large-v2",
-    "bofenghuang/whisper-small-cv11-french",
-    "bofenghuang/whisper-small-cv11-french-punct",
-    "bofenghuang/whisper-medium-cv11-french",
-    "bofenghuang/whisper-medium-cv11-french-punct",
+    "bofenghuang/whisper-medium-cv11-german-punct",
 ]
 CHUNK_LENGTH_S = 30
 MAX_NEW_TOKENS = 225
@@ -48,7 +45,7 @@ def maybe_load_cached_pipeline(model_name):
             device=device,
         )
         # set forced_decoder_ids
-        pipe.model.config.forced_decoder_ids = pipe.tokenizer.get_decoder_prompt_ids(language="fr", task="transcribe")
+        pipe.model.config.forced_decoder_ids = pipe.tokenizer.get_decoder_prompt_ids(language="de", task="transcribe")
         # limit genneration max length
         pipe.model.config.max_length = MAX_NEW_TOKENS + 1