openai 0.23.1 → 0.23.2

data/rbi/openai/models/realtime/realtime_audio_input_turn_detection.rbi

@@ -3,259 +3,320 @@
 module OpenAI
   module Models
     module Realtime
-      class RealtimeAudioInputTurnDetection < OpenAI::Internal::Type::BaseModel
-        OrHash =
+      # Configuration for turn detection, ether Server VAD or Semantic VAD. This can be
+      # set to `null` to turn off, in which case the client must manually trigger model
+      # response.
+      #
+      # Server VAD means that the model will detect the start and end of speech based on
+      # audio volume and respond at the end of user speech.
+      #
+      # Semantic VAD is more advanced and uses a turn detection model (in conjunction
+      # with VAD) to semantically estimate whether the user has finished speaking, then
+      # dynamically sets a timeout based on this probability. For example, if user audio
+      # trails off with "uhhm", the model will score a low probability of turn end and
+      # wait longer for the user to continue speaking. This can be useful for more
+      # natural conversations, but may have a higher latency.
+      module RealtimeAudioInputTurnDetection
+        extend OpenAI::Internal::Type::Union
+
+        Variants =
           T.type_alias do
             T.any(
-              OpenAI::Realtime::RealtimeAudioInputTurnDetection,
-              OpenAI::Internal::AnyHash
+              OpenAI::Realtime::RealtimeAudioInputTurnDetection::ServerVad,
+              OpenAI::Realtime::RealtimeAudioInputTurnDetection::SemanticVad
             )
           end
 
-        # Whether or not to automatically generate a response when a VAD stop event
-        # occurs.
-        sig { returns(T.nilable(T::Boolean)) }
-        attr_reader :create_response
-
-        sig { params(create_response: T::Boolean).void }
-        attr_writer :create_response
-
-        # Used only for `semantic_vad` mode. The eagerness of the model to respond. `low`
-        # will wait longer for the user to continue speaking, `high` will respond more
-        # quickly. `auto` is the default and is equivalent to `medium`. `low`, `medium`,
-        # and `high` have max timeouts of 8s, 4s, and 2s respectively.
-        sig do
-          returns(
-            T.nilable(
-              OpenAI::Realtime::RealtimeAudioInputTurnDetection::Eagerness::OrSymbol
-            )
-          )
-        end
-        attr_reader :eagerness
-
-        sig do
-          params(
-            eagerness:
-              OpenAI::Realtime::RealtimeAudioInputTurnDetection::Eagerness::OrSymbol
-          ).void
-        end
-        attr_writer :eagerness
-
-        # Optional idle timeout after which turn detection will auto-timeout when no
-        # additional audio is received and emits a `timeout_triggered` event.
-        sig { returns(T.nilable(Integer)) }
-        attr_accessor :idle_timeout_ms
-
-        # Whether or not to automatically interrupt any ongoing response with output to
-        # the default conversation (i.e. `conversation` of `auto`) when a VAD start event
-        # occurs.
-        sig { returns(T.nilable(T::Boolean)) }
-        attr_reader :interrupt_response
-
-        sig { params(interrupt_response: T::Boolean).void }
-        attr_writer :interrupt_response
-
-        # Used only for `server_vad` mode. Amount of audio to include before the VAD
-        # detected speech (in milliseconds). Defaults to 300ms.
-        sig { returns(T.nilable(Integer)) }
-        attr_reader :prefix_padding_ms
-
-        sig { params(prefix_padding_ms: Integer).void }
-        attr_writer :prefix_padding_ms
-
-        # Used only for `server_vad` mode. Duration of silence to detect speech stop (in
-        # milliseconds). Defaults to 500ms. With shorter values the model will respond
-        # more quickly, but may jump in on short pauses from the user.
-        sig { returns(T.nilable(Integer)) }
-        attr_reader :silence_duration_ms
+        class ServerVad < OpenAI::Internal::Type::BaseModel
+          OrHash =
+            T.type_alias do
+              T.any(
+                OpenAI::Realtime::RealtimeAudioInputTurnDetection::ServerVad,
+                OpenAI::Internal::AnyHash
+              )
+            end
 
-        sig { params(silence_duration_ms: Integer).void }
-        attr_writer :silence_duration_ms
+          # Type of turn detection, `server_vad` to turn on simple Server VAD.
+          sig { returns(Symbol) }
+          attr_accessor :type
 
-        # Used only for `server_vad` mode. Activation threshold for VAD (0.0 to 1.0), this
-        # defaults to 0.5. A higher threshold will require louder audio to activate the
-        # model, and thus might perform better in noisy environments.
-        sig { returns(T.nilable(Float)) }
-        attr_reader :threshold
+          # Whether or not to automatically generate a response when a VAD stop event
+          # occurs.
+          sig { returns(T.nilable(T::Boolean)) }
+          attr_reader :create_response
 
-        sig { params(threshold: Float).void }
-        attr_writer :threshold
+          sig { params(create_response: T::Boolean).void }
+          attr_writer :create_response
 
-        # Type of turn detection.
-        sig do
-          returns(
-            T.nilable(
-              OpenAI::Realtime::RealtimeAudioInputTurnDetection::Type::OrSymbol
-            )
-          )
-        end
-        attr_reader :type
+          # Optional timeout after which a model response will be triggered automatically.
+          # This is useful for situations in which a long pause from the user is unexpected,
+          # such as a phone call. The model will effectively prompt the user to continue the
+          # conversation based on the current context.
+          #
+          # The timeout value will be applied after the last model response's audio has
+          # finished playing, i.e. it's set to the `response.done` time plus audio playback
+          # duration.
+          #
+          # An `input_audio_buffer.timeout_triggered` event (plus events associated with the
+          # Response) will be emitted when the timeout is reached. Idle timeout is currently
+          # only supported for `server_vad` mode.
+          sig { returns(T.nilable(Integer)) }
+          attr_accessor :idle_timeout_ms
 
-        sig do
-          params(
-            type:
-              OpenAI::Realtime::RealtimeAudioInputTurnDetection::Type::OrSymbol
-          ).void
-        end
-        attr_writer :type
-
-        # Configuration for turn detection, ether Server VAD or Semantic VAD. This can be
-        # set to `null` to turn off, in which case the client must manually trigger model
-        # response. Server VAD means that the model will detect the start and end of
-        # speech based on audio volume and respond at the end of user speech. Semantic VAD
-        # is more advanced and uses a turn detection model (in conjunction with VAD) to
-        # semantically estimate whether the user has finished speaking, then dynamically
-        # sets a timeout based on this probability. For example, if user audio trails off
-        # with "uhhm", the model will score a low probability of turn end and wait longer
-        # for the user to continue speaking. This can be useful for more natural
-        # conversations, but may have a higher latency.
-        sig do
-          params(
-            create_response: T::Boolean,
-            eagerness:
-              OpenAI::Realtime::RealtimeAudioInputTurnDetection::Eagerness::OrSymbol,
-            idle_timeout_ms: T.nilable(Integer),
-            interrupt_response: T::Boolean,
-            prefix_padding_ms: Integer,
-            silence_duration_ms: Integer,
-            threshold: Float,
-            type:
-              OpenAI::Realtime::RealtimeAudioInputTurnDetection::Type::OrSymbol
-          ).returns(T.attached_class)
-        end
-        def self.new(
-          # Whether or not to automatically generate a response when a VAD stop event
-          # occurs.
-          create_response: nil,
-          # Used only for `semantic_vad` mode. The eagerness of the model to respond. `low`
-          # will wait longer for the user to continue speaking, `high` will respond more
-          # quickly. `auto` is the default and is equivalent to `medium`. `low`, `medium`,
-          # and `high` have max timeouts of 8s, 4s, and 2s respectively.
-          eagerness: nil,
-          # Optional idle timeout after which turn detection will auto-timeout when no
-          # additional audio is received and emits a `timeout_triggered` event.
-          idle_timeout_ms: nil,
           # Whether or not to automatically interrupt any ongoing response with output to
           # the default conversation (i.e. `conversation` of `auto`) when a VAD start event
           # occurs.
-          interrupt_response: nil,
+          sig { returns(T.nilable(T::Boolean)) }
+          attr_reader :interrupt_response
+
+          sig { params(interrupt_response: T::Boolean).void }
+          attr_writer :interrupt_response
+
           # Used only for `server_vad` mode. Amount of audio to include before the VAD
           # detected speech (in milliseconds). Defaults to 300ms.
-          prefix_padding_ms: nil,
+          sig { returns(T.nilable(Integer)) }
+          attr_reader :prefix_padding_ms
+
+          sig { params(prefix_padding_ms: Integer).void }
+          attr_writer :prefix_padding_ms
+
           # Used only for `server_vad` mode. Duration of silence to detect speech stop (in
           # milliseconds). Defaults to 500ms. With shorter values the model will respond
           # more quickly, but may jump in on short pauses from the user.
-          silence_duration_ms: nil,
+          sig { returns(T.nilable(Integer)) }
+          attr_reader :silence_duration_ms
+
+          sig { params(silence_duration_ms: Integer).void }
+          attr_writer :silence_duration_ms
+
           # Used only for `server_vad` mode. Activation threshold for VAD (0.0 to 1.0), this
           # defaults to 0.5. A higher threshold will require louder audio to activate the
           # model, and thus might perform better in noisy environments.
-          threshold: nil,
-          # Type of turn detection.
-          type: nil
-        )
-        end
+          sig { returns(T.nilable(Float)) }
+          attr_reader :threshold
 
-        sig do
-          override.returns(
-            {
+          sig { params(threshold: Float).void }
+          attr_writer :threshold
+
+          # Server-side voice activity detection (VAD) which flips on when user speech is
+          # detected and off after a period of silence.
+          sig do
+            params(
               create_response: T::Boolean,
-              eagerness:
-                OpenAI::Realtime::RealtimeAudioInputTurnDetection::Eagerness::OrSymbol,
               idle_timeout_ms: T.nilable(Integer),
               interrupt_response: T::Boolean,
               prefix_padding_ms: Integer,
               silence_duration_ms: Integer,
               threshold: Float,
-              type:
-                OpenAI::Realtime::RealtimeAudioInputTurnDetection::Type::OrSymbol
-            }
+              type: Symbol
+            ).returns(T.attached_class)
+          end
+          def self.new(
+            # Whether or not to automatically generate a response when a VAD stop event
+            # occurs.
+            create_response: nil,
+            # Optional timeout after which a model response will be triggered automatically.
+            # This is useful for situations in which a long pause from the user is unexpected,
+            # such as a phone call. The model will effectively prompt the user to continue the
+            # conversation based on the current context.
+            #
+            # The timeout value will be applied after the last model response's audio has
+            # finished playing, i.e. it's set to the `response.done` time plus audio playback
+            # duration.
+            #
+            # An `input_audio_buffer.timeout_triggered` event (plus events associated with the
+            # Response) will be emitted when the timeout is reached. Idle timeout is currently
+            # only supported for `server_vad` mode.
+            idle_timeout_ms: nil,
+            # Whether or not to automatically interrupt any ongoing response with output to
+            # the default conversation (i.e. `conversation` of `auto`) when a VAD start event
+            # occurs.
+            interrupt_response: nil,
+            # Used only for `server_vad` mode. Amount of audio to include before the VAD
+            # detected speech (in milliseconds). Defaults to 300ms.
+            prefix_padding_ms: nil,
+            # Used only for `server_vad` mode. Duration of silence to detect speech stop (in
+            # milliseconds). Defaults to 500ms. With shorter values the model will respond
+            # more quickly, but may jump in on short pauses from the user.
+            silence_duration_ms: nil,
+            # Used only for `server_vad` mode. Activation threshold for VAD (0.0 to 1.0), this
+            # defaults to 0.5. A higher threshold will require louder audio to activate the
+            # model, and thus might perform better in noisy environments.
+            threshold: nil,
+            # Type of turn detection, `server_vad` to turn on simple Server VAD.
+            type: :server_vad
           )
-        end
-        def to_hash
-        end
+          end
 
-        # Used only for `semantic_vad` mode. The eagerness of the model to respond. `low`
-        # will wait longer for the user to continue speaking, `high` will respond more
-        # quickly. `auto` is the default and is equivalent to `medium`. `low`, `medium`,
-        # and `high` have max timeouts of 8s, 4s, and 2s respectively.
-        module Eagerness
-          extend OpenAI::Internal::Type::Enum
+          sig do
+            override.returns(
+              {
+                type: Symbol,
+                create_response: T::Boolean,
+                idle_timeout_ms: T.nilable(Integer),
+                interrupt_response: T::Boolean,
+                prefix_padding_ms: Integer,
+                silence_duration_ms: Integer,
+                threshold: Float
+              }
+            )
+          end
+          def to_hash
+          end
+        end
 
-          TaggedSymbol =
+        class SemanticVad < OpenAI::Internal::Type::BaseModel
+          OrHash =
             T.type_alias do
-              T.all(
-                Symbol,
-                OpenAI::Realtime::RealtimeAudioInputTurnDetection::Eagerness
+              T.any(
+                OpenAI::Realtime::RealtimeAudioInputTurnDetection::SemanticVad,
+                OpenAI::Internal::AnyHash
               )
             end
-          OrSymbol = T.type_alias { T.any(Symbol, String) }
 
-          LOW =
-            T.let(
-              :low,
-              OpenAI::Realtime::RealtimeAudioInputTurnDetection::Eagerness::TaggedSymbol
-            )
-          MEDIUM =
-            T.let(
-              :medium,
-              OpenAI::Realtime::RealtimeAudioInputTurnDetection::Eagerness::TaggedSymbol
-            )
-          HIGH =
-            T.let(
-              :high,
-              OpenAI::Realtime::RealtimeAudioInputTurnDetection::Eagerness::TaggedSymbol
-            )
-          AUTO =
-            T.let(
-              :auto,
-              OpenAI::Realtime::RealtimeAudioInputTurnDetection::Eagerness::TaggedSymbol
-            )
+          # Type of turn detection, `semantic_vad` to turn on Semantic VAD.
+          sig { returns(Symbol) }
+          attr_accessor :type
+
+          # Whether or not to automatically generate a response when a VAD stop event
+          # occurs.
+          sig { returns(T.nilable(T::Boolean)) }
+          attr_reader :create_response
+
+          sig { params(create_response: T::Boolean).void }
+          attr_writer :create_response
 
+          # Used only for `semantic_vad` mode. The eagerness of the model to respond. `low`
+          # will wait longer for the user to continue speaking, `high` will respond more
+          # quickly. `auto` is the default and is equivalent to `medium`. `low`, `medium`,
+          # and `high` have max timeouts of 8s, 4s, and 2s respectively.
           sig do
-            override.returns(
-              T::Array[
-                OpenAI::Realtime::RealtimeAudioInputTurnDetection::Eagerness::TaggedSymbol
-              ]
+            returns(
+              T.nilable(
+                OpenAI::Realtime::RealtimeAudioInputTurnDetection::SemanticVad::Eagerness::OrSymbol
+              )
             )
           end
-          def self.values
+          attr_reader :eagerness
+
+          sig do
+            params(
+              eagerness:
+                OpenAI::Realtime::RealtimeAudioInputTurnDetection::SemanticVad::Eagerness::OrSymbol
+            ).void
           end
-        end
+          attr_writer :eagerness
 
-        # Type of turn detection.
-        module Type
-          extend OpenAI::Internal::Type::Enum
+          # Whether or not to automatically interrupt any ongoing response with output to
+          # the default conversation (i.e. `conversation` of `auto`) when a VAD start event
+          # occurs.
+          sig { returns(T.nilable(T::Boolean)) }
+          attr_reader :interrupt_response
 
-          TaggedSymbol =
-            T.type_alias do
-              T.all(
-                Symbol,
-                OpenAI::Realtime::RealtimeAudioInputTurnDetection::Type
-              )
-            end
-          OrSymbol = T.type_alias { T.any(Symbol, String) }
+          sig { params(interrupt_response: T::Boolean).void }
+          attr_writer :interrupt_response
 
-          SERVER_VAD =
-            T.let(
-              :server_vad,
-              OpenAI::Realtime::RealtimeAudioInputTurnDetection::Type::TaggedSymbol
-            )
-          SEMANTIC_VAD =
-            T.let(
-              :semantic_vad,
-              OpenAI::Realtime::RealtimeAudioInputTurnDetection::Type::TaggedSymbol
-            )
+          # Server-side semantic turn detection which uses a model to determine when the
+          # user has finished speaking.
+          sig do
+            params(
+              create_response: T::Boolean,
+              eagerness:
+                OpenAI::Realtime::RealtimeAudioInputTurnDetection::SemanticVad::Eagerness::OrSymbol,
+              interrupt_response: T::Boolean,
+              type: Symbol
+            ).returns(T.attached_class)
+          end
+          def self.new(
+            # Whether or not to automatically generate a response when a VAD stop event
+            # occurs.
+            create_response: nil,
+            # Used only for `semantic_vad` mode. The eagerness of the model to respond. `low`
+            # will wait longer for the user to continue speaking, `high` will respond more
+            # quickly. `auto` is the default and is equivalent to `medium`. `low`, `medium`,
+            # and `high` have max timeouts of 8s, 4s, and 2s respectively.
+            eagerness: nil,
+            # Whether or not to automatically interrupt any ongoing response with output to
+            # the default conversation (i.e. `conversation` of `auto`) when a VAD start event
+            # occurs.
+            interrupt_response: nil,
+            # Type of turn detection, `semantic_vad` to turn on Semantic VAD.
+            type: :semantic_vad
+          )
+          end
 
           sig do
             override.returns(
-              T::Array[
-                OpenAI::Realtime::RealtimeAudioInputTurnDetection::Type::TaggedSymbol
-              ]
+              {
+                type: Symbol,
+                create_response: T::Boolean,
+                eagerness:
+                  OpenAI::Realtime::RealtimeAudioInputTurnDetection::SemanticVad::Eagerness::OrSymbol,
+                interrupt_response: T::Boolean
+              }
             )
           end
-          def self.values
+          def to_hash
+          end
+
+          # Used only for `semantic_vad` mode. The eagerness of the model to respond. `low`
+          # will wait longer for the user to continue speaking, `high` will respond more
+          # quickly. `auto` is the default and is equivalent to `medium`. `low`, `medium`,
+          # and `high` have max timeouts of 8s, 4s, and 2s respectively.
+          module Eagerness
+            extend OpenAI::Internal::Type::Enum
+
+            TaggedSymbol =
+              T.type_alias do
+                T.all(
+                  Symbol,
+                  OpenAI::Realtime::RealtimeAudioInputTurnDetection::SemanticVad::Eagerness
+                )
+              end
+            OrSymbol = T.type_alias { T.any(Symbol, String) }
+
+            LOW =
+              T.let(
+                :low,
+                OpenAI::Realtime::RealtimeAudioInputTurnDetection::SemanticVad::Eagerness::TaggedSymbol
+              )
+            MEDIUM =
+              T.let(
+                :medium,
+                OpenAI::Realtime::RealtimeAudioInputTurnDetection::SemanticVad::Eagerness::TaggedSymbol
+              )
+            HIGH =
+              T.let(
+                :high,
+                OpenAI::Realtime::RealtimeAudioInputTurnDetection::SemanticVad::Eagerness::TaggedSymbol
+              )
+            AUTO =
+              T.let(
+                :auto,
+                OpenAI::Realtime::RealtimeAudioInputTurnDetection::SemanticVad::Eagerness::TaggedSymbol
+              )
+
+            sig do
+              override.returns(
+                T::Array[
+                  OpenAI::Realtime::RealtimeAudioInputTurnDetection::SemanticVad::Eagerness::TaggedSymbol
+                ]
+              )
+            end
+            def self.values
+            end
           end
         end
+
+        sig do
+          override.returns(
+            T::Array[
+              OpenAI::Realtime::RealtimeAudioInputTurnDetection::Variants
+            ]
+          )
+        end
+        def self.variants
+        end
       end
     end
   end