Spaces:
Runtime error
Runtime error
Commit
β’
bf2a8ff
1
Parent(s):
f8dd558
revert short-form changes
Browse files- app.py +21 -55
- assets/example_1.wav +2 -2
- assets/example_2.wav +2 -2
- assets/example_3.wav +0 -3
app.py
CHANGED
@@ -1,7 +1,6 @@
|
|
1 |
-
from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline
|
2 |
from transformers.utils import is_flash_attn_2_available
|
3 |
from transformers.pipelines.audio_utils import ffmpeg_read
|
4 |
-
from threading import Thread
|
5 |
import torch
|
6 |
import gradio as gr
|
7 |
import time
|
@@ -26,7 +25,6 @@ if not use_flash_attention_2:
|
|
26 |
distilled_model = distilled_model.to_bettertransformer()
|
27 |
|
28 |
processor = AutoProcessor.from_pretrained("openai/whisper-large-v2")
|
29 |
-
streamer = TextIteratorStreamer(processor.tokenizer, skip_special_tokens=True)
|
30 |
|
31 |
model.to(device)
|
32 |
distilled_model.to(device)
|
@@ -58,6 +56,7 @@ distil_pipe = pipeline(
|
|
58 |
)
|
59 |
distil_pipe_forward = distil_pipe._forward
|
60 |
|
|
|
61 |
def transcribe(inputs):
|
62 |
if inputs is None:
|
63 |
raise gr.Error("No audio file submitted! Please record or upload an audio file before submitting your request.")
|
@@ -74,65 +73,32 @@ def transcribe(inputs):
|
|
74 |
f"Got an audio of length {round(audio_length_mins, 3)} minutes."
|
75 |
)
|
76 |
|
77 |
-
|
78 |
-
inputs = {"array": inputs, "sampling_rate": pipe.feature_extractor.sampling_rate}
|
79 |
-
|
80 |
-
def _forward_distil_time(*args, **kwargs):
|
81 |
-
global distil_runtime_pipeline
|
82 |
-
start_time = time.time()
|
83 |
-
result = distil_pipe_forward(*args, **kwargs)
|
84 |
-
distil_runtime_pipeline = time.time() - start_time
|
85 |
-
distil_runtime_pipeline = round(distil_runtime_pipeline, 2)
|
86 |
-
return result
|
87 |
-
|
88 |
-
distil_pipe._forward = _forward_distil_time
|
89 |
-
distil_text = distil_pipe(inputs.copy(), batch_size=BATCH_SIZE)["text"]
|
90 |
-
yield distil_text, distil_runtime_pipeline, None, None
|
91 |
-
|
92 |
-
def _forward_time(*args, **kwargs):
|
93 |
-
global runtime_pipeline
|
94 |
-
start_time = time.time()
|
95 |
-
result = pipe_forward(*args, **kwargs)
|
96 |
-
runtime_pipeline = time.time() - start_time
|
97 |
-
runtime_pipeline = round(runtime_pipeline, 2)
|
98 |
-
return result
|
99 |
-
|
100 |
-
pipe._forward = _forward_time
|
101 |
-
text = pipe(inputs, batch_size=BATCH_SIZE)["text"]
|
102 |
-
|
103 |
-
yield distil_text, distil_runtime_pipeline, text, runtime_pipeline
|
104 |
-
|
105 |
-
else:
|
106 |
-
input_features = processor(inputs, sampling_rate=processor.feature_extractor.sampling_rate, return_tensors="pt").input_features
|
107 |
-
input_features = input_features.to(device, dtype=torch_dtype)
|
108 |
|
109 |
-
|
110 |
-
|
111 |
-
thread = Thread(target=distilled_model.generate, kwargs=generation_kwargs)
|
112 |
-
|
113 |
-
thread.start()
|
114 |
start_time = time.time()
|
115 |
-
|
116 |
-
for generated_text in streamer:
|
117 |
-
distil_text += generated_text
|
118 |
-
yield distil_text, None, None, None
|
119 |
-
|
120 |
distil_runtime = time.time() - start_time
|
121 |
distil_runtime = round(distil_runtime, 2)
|
122 |
-
|
123 |
|
124 |
-
|
|
|
|
|
125 |
|
126 |
-
|
|
|
127 |
start_time = time.time()
|
128 |
-
|
129 |
-
for generated_text in streamer:
|
130 |
-
text += generated_text
|
131 |
-
yield distil_text, distil_runtime, text, None
|
132 |
-
|
133 |
runtime = time.time() - start_time
|
134 |
runtime = round(runtime, 2)
|
135 |
-
|
|
|
|
|
|
|
|
|
|
|
136 |
|
137 |
|
138 |
if __name__ == "__main__":
|
@@ -158,7 +124,7 @@ if __name__ == "__main__":
|
|
158 |
of the <a href="https://huggingface.co/openai/whisper-large-v2"> Whisper</a> model by OpenAI. Compared to Whisper,
|
159 |
Distil-Whisper runs 6x faster with 50% fewer parameters, while performing to within 1% word error rate (WER) on
|
160 |
out-of-distribution evaluation data.</p>
|
161 |
-
|
162 |
<p>In this demo, we perform a speed comparison between Whisper and Distil-Whisper in order to test this claim.
|
163 |
Both models use the <a href="https://huggingface.co/distil-whisper/distil-large-v2#long-form-transcription"> chunked long-form transcription algorithm</a>
|
164 |
in π€ Transformers, as well as Flash Attention. To use Distil-Whisper yourself, check the code examples on the
|
@@ -181,7 +147,7 @@ if __name__ == "__main__":
|
|
181 |
)
|
182 |
gr.Markdown("## Examples")
|
183 |
gr.Examples(
|
184 |
-
[["./assets/example_1.wav"], ["./assets/example_2.wav"]
|
185 |
audio,
|
186 |
outputs=[distil_transcription, distil_runtime, transcription, runtime],
|
187 |
fn=transcribe,
|
|
|
1 |
+
from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline
|
2 |
from transformers.utils import is_flash_attn_2_available
|
3 |
from transformers.pipelines.audio_utils import ffmpeg_read
|
|
|
4 |
import torch
|
5 |
import gradio as gr
|
6 |
import time
|
|
|
25 |
distilled_model = distilled_model.to_bettertransformer()
|
26 |
|
27 |
processor = AutoProcessor.from_pretrained("openai/whisper-large-v2")
|
|
|
28 |
|
29 |
model.to(device)
|
30 |
distilled_model.to(device)
|
|
|
56 |
)
|
57 |
distil_pipe_forward = distil_pipe._forward
|
58 |
|
59 |
+
|
60 |
def transcribe(inputs):
|
61 |
if inputs is None:
|
62 |
raise gr.Error("No audio file submitted! Please record or upload an audio file before submitting your request.")
|
|
|
73 |
f"Got an audio of length {round(audio_length_mins, 3)} minutes."
|
74 |
)
|
75 |
|
76 |
+
inputs = {"array": inputs, "sampling_rate": pipe.feature_extractor.sampling_rate}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
77 |
|
78 |
+
def _forward_distil_time(*args, **kwargs):
|
79 |
+
global distil_runtime
|
|
|
|
|
|
|
80 |
start_time = time.time()
|
81 |
+
result = distil_pipe_forward(*args, **kwargs)
|
|
|
|
|
|
|
|
|
82 |
distil_runtime = time.time() - start_time
|
83 |
distil_runtime = round(distil_runtime, 2)
|
84 |
+
return result
|
85 |
|
86 |
+
distil_pipe._forward = _forward_distil_time
|
87 |
+
distil_text = distil_pipe(inputs.copy(), batch_size=BATCH_SIZE)["text"]
|
88 |
+
yield distil_text, distil_runtime, None, None, None
|
89 |
|
90 |
+
def _forward_time(*args, **kwargs):
|
91 |
+
global runtime
|
92 |
start_time = time.time()
|
93 |
+
result = pipe_forward(*args, **kwargs)
|
|
|
|
|
|
|
|
|
94 |
runtime = time.time() - start_time
|
95 |
runtime = round(runtime, 2)
|
96 |
+
return result
|
97 |
+
|
98 |
+
pipe._forward = _forward_time
|
99 |
+
text = pipe(inputs, batch_size=BATCH_SIZE)["text"]
|
100 |
+
|
101 |
+
yield distil_text, distil_runtime, text, runtime
|
102 |
|
103 |
|
104 |
if __name__ == "__main__":
|
|
|
124 |
of the <a href="https://huggingface.co/openai/whisper-large-v2"> Whisper</a> model by OpenAI. Compared to Whisper,
|
125 |
Distil-Whisper runs 6x faster with 50% fewer parameters, while performing to within 1% word error rate (WER) on
|
126 |
out-of-distribution evaluation data.</p>
|
127 |
+
|
128 |
<p>In this demo, we perform a speed comparison between Whisper and Distil-Whisper in order to test this claim.
|
129 |
Both models use the <a href="https://huggingface.co/distil-whisper/distil-large-v2#long-form-transcription"> chunked long-form transcription algorithm</a>
|
130 |
in π€ Transformers, as well as Flash Attention. To use Distil-Whisper yourself, check the code examples on the
|
|
|
147 |
)
|
148 |
gr.Markdown("## Examples")
|
149 |
gr.Examples(
|
150 |
+
[["./assets/example_1.wav"], ["./assets/example_2.wav"]],
|
151 |
audio,
|
152 |
outputs=[distil_transcription, distil_runtime, transcription, runtime],
|
153 |
fn=transcribe,
|
assets/example_1.wav
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1e938b9f81dea096ec7d3752e90afca8d370f7a461d3a08e1a559f4440ed055d
|
3 |
+
size 1963810
|
assets/example_2.wav
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:81fc0857f7fe11416ede431db713a02fdb787bbc049802fe74c791f3b44e5bf4
|
3 |
+
size 1920044
|
assets/example_3.wav
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:81fc0857f7fe11416ede431db713a02fdb787bbc049802fe74c791f3b44e5bf4
|
3 |
-
size 1920044
|
|
|
|
|
|
|
|