Merge branch 'main' of https://gitea.joylink.club/joylink/godot-psd-training into main

Assets/training_speech/correct.mp3.import

@@ -0,0 +1,19 @@
+[remap]
+
+importer="mp3"
+type="AudioStreamMP3"
+uid="uid://dlxo0ao1kf67d"
+path="res://.godot/imported/correct.mp3-a182f39b133e3af6125dca80877bfa41.mp3str"
+
+[deps]
+
+source_file="res://Assets/training_speech/correct.mp3"
+dest_files=["res://.godot/imported/correct.mp3-a182f39b133e3af6125dca80877bfa41.mp3str"]
+
+[params]
+
+loop=false
+loop_offset=0
+bpm=0
+beat_count=0
+bar_beats=4

Communication/voice_communication.gd

@@ -1,5 +1,6 @@
 extends Node
 
+## 语音识别成功信号
 signal speech_recognition_successed
 
 ## 录音效果器
@@ -7,6 +8,17 @@ var effect: AudioEffectRecord
 ## 录音捕获效果器（用于判断录音音量）
 var capture: AudioEffectCapture
 
+## 待语音识别的文本
+var targetText: String
+## 音量最小阈值
+const VolumeMin = 0.03
+## 长时间没有说话阈值
+const LongTimeNoVoice = 1
+var hasVoice = false
+var novoiceTime = 0
+
+## 语音识别成功回复音效
+var reply_correct = preload("res://Assets/training_speech/correct.mp3")
 
 func _ready():
 	# We get the index of the "Record" bus.
@@ -14,6 +26,7 @@ func _ready():
 	# And use it to retrieve its first effect, which has been defined
 	# as an "AudioEffectRecord" resource.
 	effect = AudioServer.get_bus_effect(idx, 0)
+	# 音频数据捕获，用于判断录音音量从而判断是否有声音输入
 	capture = AudioServer.get_bus_effect(idx, 1)
 
 ## 启动录音
@@ -28,8 +41,11 @@ func stopRecord():
 	if effect.is_recording_active():
 		effect.set_recording_active(false)
 
-var audio_sd = preload("res://Assets/training_speech/sd.mp3")
-var audio_zlgzmhfzc = preload("res://Assets/training_speech/zlgzmhfzc.mp3")
+
+## 重启录音
+func restartRecord():
+	effect.set_recording_active(false)
+	effect.set_recording_active(true)
 
 
 ## 播放回复
@@ -52,58 +68,49 @@ func play_reply(reply):
 
 ## 录音并语音识别检查
 ## PS: 是协程函数，外部如果关心结果需await
-func speech_record_check(keywords):
+func speech_record_check(text: String):
+	print("录音采样频率: ", AudioServer.get_mix_rate())
+	targetText = text
 	startRecord()
 	$Timer.start()
 	await speech_recognition_successed
+	print("识别成功，结束")
 	$Timer.stop()
 	stopRecord()
 
-## 长时间没有说话阈值
-const VolumeMin = 0.05
-const LongTimeNoVoice = 1.5
-var novoiceTime = 0
 
+## 定时处理录音并识别的逻辑
 func _on_timer_timeout():
 	if effect.is_recording_active():
-		print(capture.get_buffer_length_frames(), ", ", capture.get_discarded_frames(), ", ", capture.get_frames_available())
 		var buf = capture.get_buffer(capture.get_frames_available())
-		var hasvoice = false
+		var soundDetected = false
 		for vec in buf:
 			if vec.x > VolumeMin or vec.y > VolumeMin:
 				#print("Left channel volume = ", vec.x, ", Right volume = ", vec.y)
-				hasvoice = true
-		if hasvoice:
-			if novoiceTime > LongTimeNoVoice:
-				novoiceTime = 0
-				## 重新开始录音
-				stopRecord()
-				startRecord()
+				soundDetected = true
+		# 检测到声音处理
+		if soundDetected:
+			hasVoice = true
+			novoiceTime = 0
+		# 未检测到声音处理
 		else:
 			novoiceTime += $Timer.wait_time
-			if novoiceTime >= LongTimeNoVoice:
-				print("长时间没有说话: ", novoiceTime)
+			if hasVoice and novoiceTime >= LongTimeNoVoice:
 				var rcd = effect.get_recording()
-				stopRecord()
-				await play_reply(rcd)
+				if rcd == null:
+					return
+				print("音频时长: ", rcd.get_length())
+				restartRecord()
+				#await play_reply(rcd)
 				if rcd.get_length() > 5:
+					await play_reply(reply_correct)
 					speech_recognition_successed.emit()
-				else:
-					startRecord()
-	#if effect.is_recording_active():
-		#recording = effect.get_recording()
-		#print(recording.data.size(), ", ", recording.format, ", ", recording.loop_mode, ", ", recording.get_length())
-		#stopRecord()
-		#var player = AudioStreamPlayer.new()
-		#add_child(player)
-		#player.stream = recording
-		#player.play()
-		#player.finished.connect( _on_audio_play_finished.bind(player))
-		#startRecord()
+				#else:
+					#startRecord()
+				hasVoice = false
+			# 长时间无语音输入，重启录音
+			if novoiceTime >= LongTimeNoVoice:
+				print("长时间无声音，重启录音")
+				restartRecord()
+				novoiceTime = 0
 
-
-func _on_audio_play_finished(player):
-	print(player, "播放完成")
-	player.stop()
-	player.queue_free()
-	pass

project.godot

@@ -22,8 +22,8 @@ driver/mix_rate.web=16000
 
 [display]
 
-window/size/viewport_width=1920
-window/size/viewport_height=1080
+window/size/viewport_width=1600
+window/size/viewport_height=900
 window/size/initial_position_type=0
 window/stretch/mode="viewport"
 

trainings/training1.gd

@@ -16,6 +16,4 @@ func _ready():
 
 func _on_training_desc_dialog_start_training():
 	await $VoiceCommunication.play_reply(a_xt_pslgbztmwx)
-	#await $VoiceCommunication.play_reply(a_zby_ztmjczc)
-	#await $VoiceCommunication.play_reply(a_sd)
-	await $VoiceCommunication.speech_record_check([])
+	await $VoiceCommunication.speech_record_check("值班员，上行2号门关门故障，使用LCB关闭站台门")

trainings/training1.tscn

@@ -10,7 +10,6 @@ trainingName = "站台单档滑动门关门故障处置"
 trainingDesc = "列车关门作业，突发单个站台门未正常关闭。"
 
 [node name="TrainingDescDialog" parent="." instance=ExtResource("2_oc00d")]
-size = Vector2i(400, 300)
 
 [node name="VoiceCommunication" parent="." instance=ExtResource("3_ceo1a")]