Update README.md

README.md

@@ -1,4 +1,5 @@
----WARNING--- this is the converted CrisperWhisper model into CTranslate2 to be compatible with faster_whisper framework. Since faster whisper uses a slightly different implementation for the DTW attention matrix and output format we currently do not guarantee the same timestamp accuracy as for the transformers implementation. The transcription accuracy and filler detection should work as expected. 
+---WARNING--- this is the converted CrisperWhisper model into CTranslate2 to be compatible with [faster whisper](https://github.com/SYSTRAN/faster-whisper) framework. However, due to the different implementation of the timestamp calculation in faster whisper or more precisely [CTranslate2](https://github.com/OpenNMT/CTranslate2/) we do not guarantee the same timestamp accuracy as with the transformers implementation. The transcription accuracy and filler detection should work as expected. 
+
 
 # CrisperWhisper
 
@@ -21,7 +22,7 @@
     - [Transcription Performance](#transcription-performance)
     - [Segmentation Performance](#segmentation-performance)
 - [Usage](#2-usage)
-  - [with transformers](#21-usage-with-🤗-transformers)
+  - [with transformers](#21-usage-with-faster-whisper)
 - [How?](#3-How?)
 
 
@@ -87,79 +88,31 @@ The following table uses the metrics as defined in the paper. For this table we
 
 Here's how to use CrisperWhisper in your Python scripts:
 
-### 2.1 Usage with 🤗 transformers
 
+### 2.1 Usage with faster whisper
 
-```python
-import os
-import sys
-import torch
+We also provide a converted model to be compatible with [faster whisper](https://github.com/SYSTRAN/faster-whisper). However, due to the different implementation of the timestamp calculation in faster whisper or more precisely [CTranslate2](https://github.com/OpenNMT/CTranslate2/) the timestamp accuracy can not be guaranteed. 
 
+```python
+from faster_whisper import WhisperModel
 from datasets import load_dataset
-from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline
-
-def adjust_pauses_for_hf_pipeline_output(pipeline_output, split_threshold=0.12):
-    """
-    Adjust pause timings by distributing pauses up to the threshold evenly between adjacent words.
-    """
-
-    adjusted_chunks = pipeline_output["chunks"].copy()
-
-    for i in range(len(adjusted_chunks) - 1):
-        current_chunk = adjusted_chunks[i]
-        next_chunk = adjusted_chunks[i + 1]
-
-        current_start, current_end = current_chunk["timestamp"]
-        next_start, next_end = next_chunk["timestamp"]
-        pause_duration = next_start - current_end
-
-        if pause_duration > 0:
-            if pause_duration > split_threshold:
-                distribute = split_threshold / 2
-            else:
-                distribute = pause_duration / 2
-
-            # Adjust current chunk end time
-            adjusted_chunks[i]["timestamp"] = (current_start, current_end + distribute)
-
-            # Adjust next chunk start time
-            adjusted_chunks[i + 1]["timestamp"] = (next_start - distribute, next_end)
-    pipeline_output["chunks"] = adjusted_chunks
-
-    return pipeline_output
+faster_whisper_model = '/home/azureuser/data2/models/faster_crisper_whisper_verbatim_timestamp_finetuned_de_en_swiss'
 
+# Initialize the Whisper model
 
 device = "cuda:0" if torch.cuda.is_available() else "cpu"
-torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32
-
-model_id = "nyrahealth/CrisperWhisper"
-
-model = AutoModelForSpeechSeq2Seq.from_pretrained(
-    model_id, torch_dtype=torch_dtype, low_cpu_mem_usage=True, use_safetensors=True
-)
-model.to(device)
-
-processor = AutoProcessor.from_pretrained(model_id)
-
-pipe = pipeline(
-    "automatic-speech-recognition",
-    model=model,
-    tokenizer=processor.tokenizer,
-    feature_extractor=processor.feature_extractor,
-    chunk_length_s=30,
-    batch_size=16,
-    return_timestamps='word',
-    torch_dtype=torch_dtype,
-    device=device,
-)
-
+torch_dtype = "float16" if torch.cuda.is_available() else "float32"
+model = WhisperModel(faster_whisper_model, device=device, compute_type="float32")
 dataset = load_dataset("distil-whisper/librispeech_long", "clean", split="validation")
 sample = dataset[0]["audio"]
-hf_pipeline_output = pipe(sample)
-crisper_whisper_result = adjust_pauses_for_hf_pipeline_output(hf_pipeline_output)
-print(crisper_whisper_result)
+
+segments, info = model.transcribe(sample['array'], beam_size=1, language='en', word_timestamps = True, without_timestamps= True)
+
+for segment in segments:
+    print(segment)
 ```
 
+
 read more about the reasoning behind the pause distribution logic in our paper.
 
 ## 3. How?