Update to distil-whisper/distil-large-v3

docs/FAQ.md

@@ -225,7 +225,7 @@ export gptmore=0
 export visionmodels=1
 export enforce_h2ogpt_ui_key=False
 export top_k_docs=10
-export asr_model="distil-whisper/distil-large-v2"   #"openai/whisper-large-v3"
+export asr_model="distil-whisper/distil-large-v3"   #"openai/whisper-large-v3"
 export tts_model='microsoft/speecht5_tts'
 #export tts_model=''
 export max_max_new_tokens=8192
@@ -295,7 +295,7 @@ export gptmore=0
 export visionmodels=1
 export enforce_h2ogpt_ui_key=False
 export top_k_docs=-1
-#export asr_model="distil-whisper/distil-large-v2" #"openai/whisper-large-v3"
+#export asr_model="distil-whisper/distil-large-v3" #"openai/whisper-large-v3"
 export asr_model="openai/whisper-large-v3"
 export tts_model="tts_models/multilingual/multi-dataset/xtts_v2"
 export max_max_new_tokens=8192
@@ -1015,7 +1015,7 @@ For Twitter, one can right-click on Twitter video, copy video address, then past
 
 ### Faster ASR
 
-For fast performance, one can use `distil-whisper/distil-large-v2` or `distil-whisper/distil-large-v3` as the model, which is about 10x faster for similar accuracy.
+For fast performance, one can use `distil-whisper/distil-large-v3` or `distil-whisper/distil-large-v3` as the model, which is about 10x faster for similar accuracy.
 
 In addition, `faster_whisper` package can be used if using large v2 or v3, which is about 4x faster and 2x less memory for similar accuracy.