Add without_timestamps parameter

faster_whisper/transcribe.py

@@ -33,6 +33,7 @@ class TranscriptionOptions(
             "condition_on_previous_text",
             "temperatures",
             "initial_prompt",
+            "without_timestamps",
         ),
     )
 ):
@@ -101,6 +102,7 @@ class WhisperModel:
         no_speech_threshold=0.6,
         condition_on_previous_text=True,
         initial_prompt=None,
+        without_timestamps=False,
     ):
         """Transcribes an input file.
 
@@ -126,6 +128,7 @@ class WhisperModel:
             windows, but the model becomes less prone to getting stuck in a failure loop,
             such as repetition looping or timestamps going out of sync.
           initial_prompt: Optional text to provide as a prompt for the first window.
+          without_timestamps: Only sample text tokens.
 
         Returns:
           A tuple with:
@@ -159,6 +162,7 @@ class WhisperModel:
                 temperature if isinstance(temperature, (list, tuple)) else [temperature]
             ),
             initial_prompt=initial_prompt,
+            without_timestamps=without_timestamps,
         )
 
         segments = self.generate_segments(features, language, options)
@@ -205,7 +209,9 @@ class WhisperModel:
             segment_duration = segment.shape[-1] * self.feature_extractor.time_per_frame
 
             previous_tokens = all_tokens[prompt_reset_since:]
-            prompt = self.get_prompt(language, previous_tokens)
+            prompt = self.get_prompt(
+                language, previous_tokens, without_timestamps=options.without_timestamps
+            )
             result, temperature = self.generate_with_fallback(segment, prompt, options)
 
             if (
@@ -317,7 +323,7 @@ class WhisperModel:
 
         return result, final_temperature
 
-    def get_prompt(self, language, previous_tokens):
+    def get_prompt(self, language, previous_tokens, without_timestamps=False):
         prompt = []
 
         if previous_tokens:
@@ -330,6 +336,9 @@ class WhisperModel:
             self.tokenizer.token_to_id("<|transcribe|>"),
         ]
 
+        if without_timestamps:
+            prompt.append(self.tokenizer.token_to_id("<|notimestamps|>"))
+
         return prompt
 
     def get_segment(self, features, offset=0):