Add FSC code

README.md

@@ -1,5 +1,5 @@
 ---
-title: ESPnet2 TTS
+title: ESPnet2 SLU
 emoji: 📈
 colorFrom: green
 colorTo: green

app.py

@@ -10,95 +10,34 @@ from espnet2.bin.asr_inference import Speech2Text
 # tagen = 'kan-bayashi/ljspeech_vits' 
 # vocoder_tagen = "none" 
 
-speech2text = Speech2Text.from_pretrained(
+speech2text_slurp = Speech2Text.from_pretrained(
     asr_train_config="slurp/config.yaml",
     asr_model_file="slurp/valid.acc.ave_10best.pth",
     # Decoding parameters are not included in the model file
     nbest=1
 )
-# Confirm the sampling rate is equal to that of the training corpus.
-# If not, you need to resample the audio data before inputting to speech2text
-# speech, rate = soundfile.read("audio--1504190171-headset.flac")
-# nbests = speech2text(speech)
 
-# text, *_ = nbests[0]
-# print(text)
-# exit()
-
-# text2speechen = Text2Speech.from_pretrained(
-#     model_tag=str_or_none(tagen),
-#     vocoder_tag=str_or_none(vocoder_tagen),
-#     device="cpu",
-#     # Only for Tacotron 2 & Transformer
-#     threshold=0.5,
-#     # Only for Tacotron 2
-#     minlenratio=0.0,
-#     maxlenratio=10.0,
-#     use_att_constraint=False,
-#     backward_window=1,
-#     forward_window=3,
-#     # Only for FastSpeech & FastSpeech2 & VITS
-#     speed_control_alpha=1.0,
-#     # Only for VITS
-#     noise_scale=0.333,
-#     noise_scale_dur=0.333,
-# )
-
-
-# tagjp = 'kan-bayashi/jsut_full_band_vits_prosody' 
-# vocoder_tagjp = 'none'
-
-# text2speechjp = Text2Speech.from_pretrained(
-#     model_tag=str_or_none(tagjp),
-#     vocoder_tag=str_or_none(vocoder_tagjp),
-#     device="cpu",
-#     # Only for Tacotron 2 & Transformer
-#     threshold=0.5,
-#     # Only for Tacotron 2
-#     minlenratio=0.0,
-#     maxlenratio=10.0,
-#     use_att_constraint=False,
-#     backward_window=1,
-#     forward_window=3,
-#     # Only for FastSpeech & FastSpeech2 & VITS
-#     speed_control_alpha=1.0,
-#     # Only for VITS
-#     noise_scale=0.333,
-#     noise_scale_dur=0.333,
-# )
-
-# tagch = 'kan-bayashi/csmsc_full_band_vits'
-# vocoder_tagch = "none" 
-
-# text2speechch = Text2Speech.from_pretrained(
-#     model_tag=str_or_none(tagch),
-#     vocoder_tag=str_or_none(vocoder_tagch),
-#     device="cpu",
-#     # Only for Tacotron 2 & Transformer
-#     threshold=0.5,
-#     # Only for Tacotron 2
-#     minlenratio=0.0,
-#     maxlenratio=10.0,
-#     use_att_constraint=False,
-#     backward_window=1,
-#     forward_window=3,
-#     # Only for FastSpeech & FastSpeech2 & VITS
-#     speed_control_alpha=1.0,
-#     # Only for VITS
-#     noise_scale=0.333,
-#     noise_scale_dur=0.333,
-# )
+speech2text_fsc = Speech2Text.from_pretrained(
+    asr_train_config="fsc/config.yaml",
+    asr_model_file="fsc/valid.acc.ave_5best.pth",
+    # Decoding parameters are not included in the model file
+    nbest=1
+)
 
-def inference(wav,lang):
+def inference(wav,data):
   with torch.no_grad():
-      if lang == "english":
+      if data == "english_slurp":
           speech, rate = soundfile.read(wav.name)
-          nbests = speech2text(speech)
+          nbests = speech2text_slurp(speech)
           text, *_ = nbests[0]
           intent=text.split(" ")[0]
           scenario=intent.split("_")[0]
           action=intent.split("_")[1]
           text="{scenario: "+scenario+", action: "+action+"}"
+      elif data == "english_fsc":
+          speech, rate = soundfile.read(wav.name)
+          nbests = speech2text_fsc(speech)
+          text, *_ = nbests[0]
       # if lang == "chinese":
       #     wav = text2speechch(text)["wav"]
       #     scipy.io.wavfile.write("out.wav",text2speechch.fs , wav.view(-1).cpu().numpy())
@@ -107,15 +46,15 @@ def inference(wav,lang):
       #     scipy.io.wavfile.write("out.wav",text2speechjp.fs , wav.view(-1).cpu().numpy())
   return  text
 title = "ESPnet2-SLU"
-description = "Gradio demo for ESPnet2-SLU: Advancing Spoken Language Understanding through ESPnet. To use it, simply record your audio. Read more at the links below."
+description = "Gradio demo for ESPnet2-SLU: Advancing Spoken Language Understanding through ESPnet. To use it, simply record your audio or click one of the examples to load them. Read more at the links below."
 article = "<p style='text-align: center'><a href='https://github.com/espnet/espnet' target='_blank'>Github Repo</a></p>"
 
-examples=[['audio_slurp.flac',"english"]]
+examples=[['audio_slurp.flac',"english_slurp"],['audio_fsc.wav',"english_fsc"]]
 
 # gr.inputs.Textbox(label="input text",lines=10),gr.inputs.Radio(choices=["english"], type="value", default="english", label="language")
 gr.Interface(
     inference, 
-    [gr.inputs.Audio(label="input audio",source = "microphone", type="file"),gr.inputs.Radio(choices=["english"], type="value", default="english", label="language")], 
+    [gr.inputs.Audio(label="input audio",source = "microphone", type="file"),gr.inputs.Radio(choices=["english_slurp","english_fsc"], type="value", default="english_slurp", label="Dataset")], 
     gr.outputs.Textbox(type="str", label="Output"),
     title=title,
     description=description,

fsc/config.yaml

@@ -0,0 +1,355 @@
+config: conf/tuning/train_asr_hubert_transformer_adam_specaug.yaml
+print_config: false
+log_level: INFO
+dry_run: false
+iterator_type: sequence
+output_dir: exp/asr_train_asr_hubert_transformer_adam_specaug_raw_en_word
+ngpu: 1
+seed: 0
+num_workers: 1
+num_att_plot: 3
+dist_backend: nccl
+dist_init_method: env://
+dist_world_size: null
+dist_rank: null
+local_rank: 0
+dist_master_addr: null
+dist_master_port: null
+dist_launcher: null
+multiprocessing_distributed: false
+unused_parameters: false
+sharded_ddp: false
+cudnn_enabled: true
+cudnn_benchmark: false
+cudnn_deterministic: true
+collect_stats: false
+write_collected_feats: false
+max_epoch: 200
+patience: null
+val_scheduler_criterion:
+- valid
+- loss
+early_stopping_criterion:
+- valid
+- loss
+- min
+best_model_criterion:
+-   - train
+    - loss
+    - min
+-   - valid
+    - loss
+    - min
+-   - train
+    - acc
+    - max
+-   - valid
+    - acc
+    - max
+keep_nbest_models: 5
+grad_clip: 5.0
+grad_clip_type: 2.0
+grad_noise: false
+accum_grad: 1
+no_forward_run: false
+resume: true
+train_dtype: float32
+use_amp: false
+log_interval: null
+use_tensorboard: true
+use_wandb: false
+wandb_project: null
+wandb_id: null
+wandb_entity: null
+wandb_name: null
+wandb_model_log_interval: -1
+detect_anomaly: false
+pretrain_path: null
+init_param: []
+ignore_init_mismatch: false
+freeze_param:
+- frontend.upstream
+num_iters_per_epoch: null
+batch_size: 20
+valid_batch_size: null
+batch_bins: 1000000
+valid_batch_bins: null
+train_shape_file:
+- exp/asr_stats_raw_en_word/train/speech_shape
+- exp/asr_stats_raw_en_word/train/text_shape.word
+valid_shape_file:
+- exp/asr_stats_raw_en_word/valid/speech_shape
+- exp/asr_stats_raw_en_word/valid/text_shape.word
+batch_type: folded
+valid_batch_type: null
+fold_length:
+- 80000
+- 150
+sort_in_batch: descending
+sort_batch: descending
+multiple_iterator: false
+chunk_length: 500
+chunk_shift_ratio: 0.5
+num_cache_chunks: 1024
+train_data_path_and_name_and_type:
+-   - dump/raw/train/wav.scp
+    - speech
+    - sound
+-   - dump/raw/train/text
+    - text
+    - text
+valid_data_path_and_name_and_type:
+-   - dump/raw/valid/wav.scp
+    - speech
+    - sound
+-   - dump/raw/valid/text
+    - text
+    - text
+allow_variable_data_keys: false
+max_cache_size: 0.0
+max_cache_fd: 32
+valid_max_cache_size: null
+optim: adam
+optim_conf:
+    lr: 0.0002
+scheduler: warmuplr
+scheduler_conf:
+    warmup_steps: 25000
+token_list:
+- <blank>
+- <unk>
+- the
+- Turn
+- in
+- lights
+- 'on'
+- up
+- down
+- temperature
+- heat
+- 'off'
+- Switch
+- increase_volume_none
+- kitchen
+- language
+- decrease_volume_none
+- bedroom
+- washroom
+- volume
+- my
+- to
+- bathroom
+- Decrease
+- increase_heat_washroom
+- decrease_heat_washroom
+- Increase
+- music
+- heating
+- Bring
+- increase_heat_none
+- decrease_heat_none
+- me
+- change_language_none_none
+- activate_lights_washroom
+- Set
+- Lights
+- activate_lights_kitchen
+- I
+- activate_music_none
+- too
+- it
+- increase_heat_bedroom
+- decrease_heat_bedroom
+- sound
+- increase_heat_kitchen
+- decrease_heat_kitchen
+- deactivate_music_none
+- lamp
+- Make
+- deactivate_lights_bedroom
+- deactivate_lights_kitchen
+- bring_newspaper_none
+- newspaper
+- activate_lights_bedroom
+- bring_socks_none
+- socks
+- bring_shoes_none
+- shoes
+- need
+- Volume
+- activate_lights_none
+- deactivate_lights_none
+- bring_juice_none
+- juice
+- deactivate_lights_washroom
+- change_language_Chinese_none
+- deactivate_lamp_none
+- activate_lamp_none
+- Kitchen
+- turn
+- some
+- Could
+- you
+- Bedroom
+- Go
+- get
+- Washroom
+- Chinese
+- phone's
+- change_language_English_none
+- Get
+- change_language_Korean_none
+- OK
+- now
+- switch
+- main
+- change_language_German_none
+- practice
+- Louder
+- Stop
+- loud
+- increase
+- Play
+- hear
+- Change
+- quiet
+- Bathroom
+- Fetch
+- Korean
+- English
+- German
+- Pause
+- Lamp
+- Resume
+- louder
+- Heat
+- audio
+- Its
+- loud,
+- heating?
+- Far
+- a
+- different
+- please?
+- decrease
+- Too
+- settings
+- Put
+- Start
+- Quieter
+- please
+- Thats
+- softer
+- max
+- mute
+- lower
+- phone
+- couldn't
+- anything,
+- Reduce
+- this,
+- More
+- That's
+- Lower
+- levels
+- Use
+- hotter
+- languages
+- Allow
+- can't
+- that
+- Less
+- system
+- cooler
+- This
+- video
+- is
+- low,
+- device
+- Chinese.
+- quieter
+- English.
+- Language
+- Open
+- German.
+- Korean.
+- <sos/eos>
+init: null
+input_size: null
+ctc_conf:
+    dropout_rate: 0.0
+    ctc_type: builtin
+    reduce: true
+    ignore_nan_grad: true
+model_conf:
+    ctc_weight: 0.3
+    lsm_weight: 0.1
+    length_normalized_loss: false
+    extract_feats_in_collect_stats: false
+use_preprocessor: true
+token_type: word
+bpemodel: null
+non_linguistic_symbols: null
+cleaner: null
+g2p: null
+speech_volume_normalize: null
+rir_scp: null
+rir_apply_prob: 1.0
+noise_scp: null
+noise_apply_prob: 1.0
+noise_db_range: '13_15'
+frontend: s3prl
+frontend_conf:
+    frontend_conf:
+        upstream: hubert_large_ll60k
+    download_dir: ./hub
+    multilayer_feature: true
+    fs: 16k
+specaug: specaug
+specaug_conf:
+    apply_time_warp: true
+    time_warp_window: 5
+    time_warp_mode: bicubic
+    apply_freq_mask: true
+    freq_mask_width_range:
+    - 0
+    - 30
+    num_freq_mask: 2
+    apply_time_mask: true
+    time_mask_width_range:
+    - 0
+    - 40
+    num_time_mask: 2
+normalize: utterance_mvn
+normalize_conf: {}
+preencoder: linear
+preencoder_conf:
+    input_size: 1024
+    output_size: 80
+encoder: transformer
+encoder_conf:
+    output_size: 256
+    attention_heads: 4
+    linear_units: 2048
+    num_blocks: 12
+    dropout_rate: 0.1
+    positional_dropout_rate: 0.1
+    attention_dropout_rate: 0.0
+    input_layer: conv2d
+    normalize_before: true
+postencoder: null
+postencoder_conf: {}
+decoder: transformer
+decoder_conf:
+    attention_heads: 4
+    linear_units: 2048
+    num_blocks: 6
+    dropout_rate: 0.1
+    positional_dropout_rate: 0.1
+    self_attention_dropout_rate: 0.0
+    src_attention_dropout_rate: 0.0
+required:
+- output_dir
+- token_list
+version: 0.10.3a2
+distributed: false

fsc/valid.acc.ave_5best.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d2adc40ef8aaace766e9aa307cc49e78419a516a75fa58b3eabecc9e857fc5a4
+size 1375946815