Spaces:
Running
Running
Irpan
commited on
Commit
·
c492cbb
1
Parent(s):
448bf1b
asr
Browse files
app.py
CHANGED
@@ -14,7 +14,7 @@ mms_transcribe = gr.Interface(
|
|
14 |
gr.Dropdown(
|
15 |
choices=[model for model in asr.models_info],
|
16 |
label="Select a Model",
|
17 |
-
value="
|
18 |
interactive=True
|
19 |
),
|
20 |
],
|
@@ -37,7 +37,7 @@ mms_synthesize = gr.Interface(
|
|
37 |
gr.Dropdown(
|
38 |
choices=[model for model in tts.models_info],
|
39 |
label="Select a Model",
|
40 |
-
value="
|
41 |
interactive=True
|
42 |
)
|
43 |
],
|
|
|
14 |
gr.Dropdown(
|
15 |
choices=[model for model in asr.models_info],
|
16 |
label="Select a Model",
|
17 |
+
value="Ixxan-FineTuned-MMS",
|
18 |
interactive=True
|
19 |
),
|
20 |
],
|
|
|
37 |
gr.Dropdown(
|
38 |
choices=[model for model in tts.models_info],
|
39 |
label="Select a Model",
|
40 |
+
value="Ixxan-FineTuned-MMS",
|
41 |
interactive=True
|
42 |
)
|
43 |
],
|
asr.py
CHANGED
@@ -25,19 +25,13 @@ models_info = {
|
|
25 |
"ctc_model": True,
|
26 |
"arabic_script": True
|
27 |
},
|
28 |
-
"
|
29 |
-
"processor": AutoProcessor.from_pretrained("ixxan/whisper-small-thugy20"),
|
30 |
-
"model": AutoModelForSpeechSeq2Seq.from_pretrained("ixxan/whisper-small-thugy20"),
|
31 |
-
"ctc_model": False,
|
32 |
-
"arabic_script": False
|
33 |
-
},
|
34 |
-
"ixxan/whisper-small-uyghur-common-voice": {
|
35 |
"processor": AutoProcessor.from_pretrained("ixxan/whisper-small-uyghur-common-voice"),
|
36 |
"model": AutoModelForSpeechSeq2Seq.from_pretrained("ixxan/whisper-small-uyghur-common-voice"),
|
37 |
"ctc_model": False,
|
38 |
"arabic_script": False
|
39 |
},
|
40 |
-
"
|
41 |
"processor": Wav2Vec2Processor.from_pretrained("ixxan/wav2vec2-large-mms-1b-uyghur-latin", target_lang='uig-script_latin'),
|
42 |
"model": Wav2Vec2ForCTC.from_pretrained("ixxan/wav2vec2-large-mms-1b-uyghur-latin", target_lang='uig-script_latin'),
|
43 |
"ctc_model": True,
|
|
|
25 |
"ctc_model": True,
|
26 |
"arabic_script": True
|
27 |
},
|
28 |
+
"Ixxan-FineTuned-Whisper": {
|
|
|
|
|
|
|
|
|
|
|
|
|
29 |
"processor": AutoProcessor.from_pretrained("ixxan/whisper-small-uyghur-common-voice"),
|
30 |
"model": AutoModelForSpeechSeq2Seq.from_pretrained("ixxan/whisper-small-uyghur-common-voice"),
|
31 |
"ctc_model": False,
|
32 |
"arabic_script": False
|
33 |
},
|
34 |
+
"Ixxan-FineTuned-MMS": {
|
35 |
"processor": Wav2Vec2Processor.from_pretrained("ixxan/wav2vec2-large-mms-1b-uyghur-latin", target_lang='uig-script_latin'),
|
36 |
"model": Wav2Vec2ForCTC.from_pretrained("ixxan/wav2vec2-large-mms-1b-uyghur-latin", target_lang='uig-script_latin'),
|
37 |
"ctc_model": True,
|
tts.py
CHANGED
@@ -10,12 +10,17 @@ device = "cuda" if torch.cuda.is_available() else "cpu"
|
|
10 |
|
11 |
# Load processor and model
|
12 |
models_info = {
|
|
|
13 |
"Meta-MMS": {
|
14 |
"processor": AutoTokenizer.from_pretrained("facebook/mms-tts-uig-script_arabic"),
|
15 |
"model": VitsModel.from_pretrained("facebook/mms-tts-uig-script_arabic"),
|
16 |
"arabic_script": True
|
17 |
},
|
18 |
-
"
|
|
|
|
|
|
|
|
|
19 |
}
|
20 |
|
21 |
vocoder_checkpoint="parallelwavegan_male2_checkpoint/checkpoint-400000steps.pkl" ### specify vocoder path
|
|
|
10 |
|
11 |
# Load processor and model
|
12 |
models_info = {
|
13 |
+
"IS2AI-TurkicTTS": None,
|
14 |
"Meta-MMS": {
|
15 |
"processor": AutoTokenizer.from_pretrained("facebook/mms-tts-uig-script_arabic"),
|
16 |
"model": VitsModel.from_pretrained("facebook/mms-tts-uig-script_arabic"),
|
17 |
"arabic_script": True
|
18 |
},
|
19 |
+
"Ixxan-FineTuned-MMS": {
|
20 |
+
"processor": AutoTokenizer.from_pretrained("ixxan/mms-tts-uig-script_arabic-UQSpeech"),
|
21 |
+
"model": VitsModel.from_pretrained("ixxan/mms-tts-uig-script_arabic-UQSpeech"),
|
22 |
+
"arabic_script": True
|
23 |
+
}
|
24 |
}
|
25 |
|
26 |
vocoder_checkpoint="parallelwavegan_male2_checkpoint/checkpoint-400000steps.pkl" ### specify vocoder path
|