openai 0.23.0 → 0.23.1

data/rbi/openai/models/realtime/transcription_session_updated_event.rbi

@@ -22,14 +22,14 @@ module OpenAI
         # contains an ephemeral key. Default TTL for keys is 10 minutes. This property is
         # not present when a session is updated via the WebSocket API.
         sig do
-          returns(OpenAI::Realtime::RealtimeTranscriptionSessionCreateResponse)
+          returns(OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session)
         end
         attr_reader :session
 
         sig do
           params(
             session:
-              OpenAI::Realtime::RealtimeTranscriptionSessionCreateResponse::OrHash
+              OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session::OrHash
           ).void
         end
         attr_writer :session
@@ -44,7 +44,7 @@ module OpenAI
           params(
             event_id: String,
             session:
-              OpenAI::Realtime::RealtimeTranscriptionSessionCreateResponse::OrHash,
+              OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session::OrHash,
             type: Symbol
           ).returns(T.attached_class)
         end
@@ -67,13 +67,320 @@ module OpenAI
             {
               event_id: String,
               session:
-                OpenAI::Realtime::RealtimeTranscriptionSessionCreateResponse,
+                OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session,
               type: Symbol
             }
           )
         end
         def to_hash
         end
+
+        class Session < OpenAI::Internal::Type::BaseModel
+          OrHash =
+            T.type_alias do
+              T.any(
+                OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session,
+                OpenAI::Internal::AnyHash
+              )
+            end
+
+          # Ephemeral key returned by the API. Only present when the session is created on
+          # the server via REST API.
+          sig do
+            returns(
+              OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session::ClientSecret
+            )
+          end
+          attr_reader :client_secret
+
+          sig do
+            params(
+              client_secret:
+                OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session::ClientSecret::OrHash
+            ).void
+          end
+          attr_writer :client_secret
+
+          # The format of input audio. Options are `pcm16`, `g711_ulaw`, or `g711_alaw`.
+          sig { returns(T.nilable(String)) }
+          attr_reader :input_audio_format
+
+          sig { params(input_audio_format: String).void }
+          attr_writer :input_audio_format
+
+          # Configuration of the transcription model.
+          sig { returns(T.nilable(OpenAI::Realtime::AudioTranscription)) }
+          attr_reader :input_audio_transcription
+
+          sig do
+            params(
+              input_audio_transcription:
+                OpenAI::Realtime::AudioTranscription::OrHash
+            ).void
+          end
+          attr_writer :input_audio_transcription
+
+          # The set of modalities the model can respond with. To disable audio, set this to
+          # ["text"].
+          sig do
+            returns(
+              T.nilable(
+                T::Array[
+                  OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session::Modality::OrSymbol
+                ]
+              )
+            )
+          end
+          attr_reader :modalities
+
+          sig do
+            params(
+              modalities:
+                T::Array[
+                  OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session::Modality::OrSymbol
+                ]
+            ).void
+          end
+          attr_writer :modalities
+
+          # Configuration for turn detection. Can be set to `null` to turn off. 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.
+          sig do
+            returns(
+              T.nilable(
+                OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session::TurnDetection
+              )
+            )
+          end
+          attr_reader :turn_detection
+
+          sig do
+            params(
+              turn_detection:
+                OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session::TurnDetection::OrHash
+            ).void
+          end
+          attr_writer :turn_detection
+
+          # A new Realtime transcription session configuration.
+          #
+          # When a session is created on the server via REST API, the session object also
+          # contains an ephemeral key. Default TTL for keys is 10 minutes. This property is
+          # not present when a session is updated via the WebSocket API.
+          sig do
+            params(
+              client_secret:
+                OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session::ClientSecret::OrHash,
+              input_audio_format: String,
+              input_audio_transcription:
+                OpenAI::Realtime::AudioTranscription::OrHash,
+              modalities:
+                T::Array[
+                  OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session::Modality::OrSymbol
+                ],
+              turn_detection:
+                OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session::TurnDetection::OrHash
+            ).returns(T.attached_class)
+          end
+          def self.new(
+            # Ephemeral key returned by the API. Only present when the session is created on
+            # the server via REST API.
+            client_secret:,
+            # The format of input audio. Options are `pcm16`, `g711_ulaw`, or `g711_alaw`.
+            input_audio_format: nil,
+            # Configuration of the transcription model.
+            input_audio_transcription: nil,
+            # The set of modalities the model can respond with. To disable audio, set this to
+            # ["text"].
+            modalities: nil,
+            # Configuration for turn detection. Can be set to `null` to turn off. 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.
+            turn_detection: nil
+          )
+          end
+
+          sig do
+            override.returns(
+              {
+                client_secret:
+                  OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session::ClientSecret,
+                input_audio_format: String,
+                input_audio_transcription: OpenAI::Realtime::AudioTranscription,
+                modalities:
+                  T::Array[
+                    OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session::Modality::OrSymbol
+                  ],
+                turn_detection:
+                  OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session::TurnDetection
+              }
+            )
+          end
+          def to_hash
+          end
+
+          class ClientSecret < OpenAI::Internal::Type::BaseModel
+            OrHash =
+              T.type_alias do
+                T.any(
+                  OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session::ClientSecret,
+                  OpenAI::Internal::AnyHash
+                )
+              end
+
+            # Timestamp for when the token expires. Currently, all tokens expire after one
+            # minute.
+            sig { returns(Integer) }
+            attr_accessor :expires_at
+
+            # Ephemeral key usable in client environments to authenticate connections to the
+            # Realtime API. Use this in client-side environments rather than a standard API
+            # token, which should only be used server-side.
+            sig { returns(String) }
+            attr_accessor :value
+
+            # Ephemeral key returned by the API. Only present when the session is created on
+            # the server via REST API.
+            sig do
+              params(expires_at: Integer, value: String).returns(
+                T.attached_class
+              )
+            end
+            def self.new(
+              # Timestamp for when the token expires. Currently, all tokens expire after one
+              # minute.
+              expires_at:,
+              # Ephemeral key usable in client environments to authenticate connections to the
+              # Realtime API. Use this in client-side environments rather than a standard API
+              # token, which should only be used server-side.
+              value:
+            )
+            end
+
+            sig { override.returns({ expires_at: Integer, value: String }) }
+            def to_hash
+            end
+          end
+
+          module Modality
+            extend OpenAI::Internal::Type::Enum
+
+            TaggedSymbol =
+              T.type_alias do
+                T.all(
+                  Symbol,
+                  OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session::Modality
+                )
+              end
+            OrSymbol = T.type_alias { T.any(Symbol, String) }
+
+            TEXT =
+              T.let(
+                :text,
+                OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session::Modality::TaggedSymbol
+              )
+            AUDIO =
+              T.let(
+                :audio,
+                OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session::Modality::TaggedSymbol
+              )
+
+            sig do
+              override.returns(
+                T::Array[
+                  OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session::Modality::TaggedSymbol
+                ]
+              )
+            end
+            def self.values
+            end
+          end
+
+          class TurnDetection < OpenAI::Internal::Type::BaseModel
+            OrHash =
+              T.type_alias do
+                T.any(
+                  OpenAI::Realtime::TranscriptionSessionUpdatedEvent::Session::TurnDetection,
+                  OpenAI::Internal::AnyHash
+                )
+              end
+
+            # 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
+
+            # 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
+
+            sig { params(silence_duration_ms: Integer).void }
+            attr_writer :silence_duration_ms
+
+            # 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
+
+            sig { params(threshold: Float).void }
+            attr_writer :threshold
+
+            # Type of turn detection, only `server_vad` is currently supported.
+            sig { returns(T.nilable(String)) }
+            attr_reader :type
+
+            sig { params(type: String).void }
+            attr_writer :type
+
+            # Configuration for turn detection. Can be set to `null` to turn off. 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.
+            sig do
+              params(
+                prefix_padding_ms: Integer,
+                silence_duration_ms: Integer,
+                threshold: Float,
+                type: String
+              ).returns(T.attached_class)
+            end
+            def self.new(
+              # Amount of audio to include before the VAD detected speech (in milliseconds).
+              # Defaults to 300ms.
+              prefix_padding_ms: nil,
+              # 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,
+              # 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, only `server_vad` is currently supported.
+              type: nil
+            )
+            end
+
+            sig do
+              override.returns(
+                {
+                  prefix_padding_ms: Integer,
+                  silence_duration_ms: Integer,
+                  threshold: Float,
+                  type: String
+                }
+              )
+            end
+            def to_hash
+            end
+          end
+        end
       end
     end
   end

data/sig/openai/models/realtime/realtime_client_event.rbs

@@ -13,7 +13,6 @@ module OpenAI
         | OpenAI::Realtime::ResponseCancelEvent
         | OpenAI::Realtime::ResponseCreateEvent
         | OpenAI::Realtime::SessionUpdateEvent
-        | OpenAI::Realtime::TranscriptionSessionUpdate
 
       module RealtimeClientEvent
         extend OpenAI::Internal::Type::Union

data/sig/openai/models/realtime/{models.rbs → realtime_function_tool.rbs}

@@ -1,15 +1,15 @@
 module OpenAI
   module Models
     module Realtime
-      type models =
+      type realtime_function_tool =
         {
           description: String,
           name: String,
           parameters: top,
-          type: OpenAI::Models::Realtime::Models::type_
+          type: OpenAI::Models::Realtime::RealtimeFunctionTool::type_
         }
 
-      class Models < OpenAI::Internal::Type::BaseModel
+      class RealtimeFunctionTool < OpenAI::Internal::Type::BaseModel
         attr_reader description: String?
 
         def description=: (String) -> String
@@ -22,24 +22,24 @@ module OpenAI
 
         def parameters=: (top) -> top
 
-        attr_reader type: OpenAI::Models::Realtime::Models::type_?
+        attr_reader type: OpenAI::Models::Realtime::RealtimeFunctionTool::type_?
 
         def type=: (
-          OpenAI::Models::Realtime::Models::type_
-        ) -> OpenAI::Models::Realtime::Models::type_
+          OpenAI::Models::Realtime::RealtimeFunctionTool::type_
+        ) -> OpenAI::Models::Realtime::RealtimeFunctionTool::type_
 
         def initialize: (
           ?description: String,
           ?name: String,
           ?parameters: top,
-          ?type: OpenAI::Models::Realtime::Models::type_
+          ?type: OpenAI::Models::Realtime::RealtimeFunctionTool::type_
         ) -> void
 
         def to_hash: -> {
           description: String,
           name: String,
           parameters: top,
-          type: OpenAI::Models::Realtime::Models::type_
+          type: OpenAI::Models::Realtime::RealtimeFunctionTool::type_
         }
 
         type type_ = :function
@@ -49,7 +49,7 @@ module OpenAI
 
           FUNCTION: :function
 
-          def self?.values: -> ::Array[OpenAI::Models::Realtime::Models::type_]
+          def self?.values: -> ::Array[OpenAI::Models::Realtime::RealtimeFunctionTool::type_]
         end
       end
     end

data/sig/openai/models/realtime/realtime_response_create_params.rbs

@@ -134,7 +134,7 @@ module OpenAI
         end
 
         type tool =
-          OpenAI::Realtime::Models
+          OpenAI::Realtime::RealtimeFunctionTool
           | OpenAI::Realtime::RealtimeResponseCreateMcpTool
 
         module Tool

data/sig/openai/models/realtime/realtime_server_event.rbs

@@ -32,8 +32,6 @@ module OpenAI
         | OpenAI::Realtime::ResponseTextDoneEvent
         | OpenAI::Realtime::SessionCreatedEvent
         | OpenAI::Realtime::SessionUpdatedEvent
-        | OpenAI::Realtime::TranscriptionSessionUpdatedEvent
-        | OpenAI::Realtime::TranscriptionSessionCreated
         | OpenAI::Realtime::RealtimeServerEvent::OutputAudioBufferStarted
         | OpenAI::Realtime::RealtimeServerEvent::OutputAudioBufferStopped
         | OpenAI::Realtime::RealtimeServerEvent::OutputAudioBufferCleared

data/sig/openai/models/realtime/realtime_session.rbs

@@ -19,7 +19,7 @@ module OpenAI
           speed: Float,
           temperature: Float,
           tool_choice: String,
-          tools: ::Array[OpenAI::Realtime::Models],
+          tools: ::Array[OpenAI::Realtime::RealtimeFunctionTool],
           tracing: OpenAI::Models::Realtime::RealtimeSession::tracing?,
           turn_detection: OpenAI::Realtime::RealtimeSession::TurnDetection?,
           voice: OpenAI::Models::Realtime::RealtimeSession::voice
@@ -98,11 +98,11 @@ module OpenAI
 
         def tool_choice=: (String) -> String
 
-        attr_reader tools: ::Array[OpenAI::Realtime::Models]?
+        attr_reader tools: ::Array[OpenAI::Realtime::RealtimeFunctionTool]?
 
         def tools=: (
-          ::Array[OpenAI::Realtime::Models]
-        ) -> ::Array[OpenAI::Realtime::Models]
+          ::Array[OpenAI::Realtime::RealtimeFunctionTool]
+        ) -> ::Array[OpenAI::Realtime::RealtimeFunctionTool]
 
         attr_accessor tracing: OpenAI::Models::Realtime::RealtimeSession::tracing?
 
@@ -131,7 +131,7 @@ module OpenAI
           ?speed: Float,
           ?temperature: Float,
           ?tool_choice: String,
-          ?tools: ::Array[OpenAI::Realtime::Models],
+          ?tools: ::Array[OpenAI::Realtime::RealtimeFunctionTool],
           ?tracing: OpenAI::Models::Realtime::RealtimeSession::tracing?,
           ?turn_detection: OpenAI::Realtime::RealtimeSession::TurnDetection?,
           ?voice: OpenAI::Models::Realtime::RealtimeSession::voice
@@ -154,7 +154,7 @@ module OpenAI
           speed: Float,
           temperature: Float,
           tool_choice: String,
-          tools: ::Array[OpenAI::Realtime::Models],
+          tools: ::Array[OpenAI::Realtime::RealtimeFunctionTool],
           tracing: OpenAI::Models::Realtime::RealtimeSession::tracing?,
           turn_detection: OpenAI::Realtime::RealtimeSession::TurnDetection?,
           voice: OpenAI::Models::Realtime::RealtimeSession::voice

data/sig/openai/models/realtime/realtime_session_create_response.rbs

@@ -5,8 +5,9 @@ module OpenAI
     module Realtime
       type realtime_session_create_response =
         {
-          audio: OpenAI::Realtime::RealtimeSessionCreateResponse::Audio,
           client_secret: OpenAI::Realtime::RealtimeSessionClientSecret,
+          type: :realtime,
+          audio: OpenAI::Realtime::RealtimeSessionCreateResponse::Audio,
           include: ::Array[OpenAI::Models::Realtime::RealtimeSessionCreateResponse::include_],
           instructions: String,
           max_output_tokens: OpenAI::Models::Realtime::RealtimeSessionCreateResponse::max_output_tokens,
@@ -16,23 +17,20 @@ module OpenAI
           tool_choice: OpenAI::Models::Realtime::RealtimeSessionCreateResponse::tool_choice,
           tools: ::Array[OpenAI::Models::Realtime::RealtimeSessionCreateResponse::tool],
           tracing: OpenAI::Models::Realtime::RealtimeSessionCreateResponse::tracing?,
-          truncation: OpenAI::Models::Realtime::realtime_truncation,
-          type: OpenAI::Models::Realtime::RealtimeSessionCreateResponse::type_
+          truncation: OpenAI::Models::Realtime::realtime_truncation
         }
 
       class RealtimeSessionCreateResponse < OpenAI::Internal::Type::BaseModel
+        attr_accessor client_secret: OpenAI::Realtime::RealtimeSessionClientSecret
+
+        attr_accessor type: :realtime
+
         attr_reader audio: OpenAI::Realtime::RealtimeSessionCreateResponse::Audio?
 
         def audio=: (
           OpenAI::Realtime::RealtimeSessionCreateResponse::Audio
         ) -> OpenAI::Realtime::RealtimeSessionCreateResponse::Audio
 
-        attr_reader client_secret: OpenAI::Realtime::RealtimeSessionClientSecret?
-
-        def client_secret=: (
-          OpenAI::Realtime::RealtimeSessionClientSecret
-        ) -> OpenAI::Realtime::RealtimeSessionClientSecret
-
         attr_reader include: ::Array[OpenAI::Models::Realtime::RealtimeSessionCreateResponse::include_]?
 
         def include=: (
@@ -83,15 +81,9 @@ module OpenAI
           OpenAI::Models::Realtime::realtime_truncation
         ) -> OpenAI::Models::Realtime::realtime_truncation
 
-        attr_reader type: OpenAI::Models::Realtime::RealtimeSessionCreateResponse::type_?
-
-        def type=: (
-          OpenAI::Models::Realtime::RealtimeSessionCreateResponse::type_
-        ) -> OpenAI::Models::Realtime::RealtimeSessionCreateResponse::type_
-
         def initialize: (
+          client_secret: OpenAI::Realtime::RealtimeSessionClientSecret,
           ?audio: OpenAI::Realtime::RealtimeSessionCreateResponse::Audio,
-          ?client_secret: OpenAI::Realtime::RealtimeSessionClientSecret,
           ?include: ::Array[OpenAI::Models::Realtime::RealtimeSessionCreateResponse::include_],
           ?instructions: String,
           ?max_output_tokens: OpenAI::Models::Realtime::RealtimeSessionCreateResponse::max_output_tokens,
@@ -102,12 +94,13 @@ module OpenAI
           ?tools: ::Array[OpenAI::Models::Realtime::RealtimeSessionCreateResponse::tool],
           ?tracing: OpenAI::Models::Realtime::RealtimeSessionCreateResponse::tracing?,
           ?truncation: OpenAI::Models::Realtime::realtime_truncation,
-          ?type: OpenAI::Models::Realtime::RealtimeSessionCreateResponse::type_
+          ?type: :realtime
         ) -> void
 
         def to_hash: -> {
-          audio: OpenAI::Realtime::RealtimeSessionCreateResponse::Audio,
           client_secret: OpenAI::Realtime::RealtimeSessionClientSecret,
+          type: :realtime,
+          audio: OpenAI::Realtime::RealtimeSessionCreateResponse::Audio,
           include: ::Array[OpenAI::Models::Realtime::RealtimeSessionCreateResponse::include_],
           instructions: String,
           max_output_tokens: OpenAI::Models::Realtime::RealtimeSessionCreateResponse::max_output_tokens,
@@ -117,8 +110,7 @@ module OpenAI
           tool_choice: OpenAI::Models::Realtime::RealtimeSessionCreateResponse::tool_choice,
           tools: ::Array[OpenAI::Models::Realtime::RealtimeSessionCreateResponse::tool],
           tracing: OpenAI::Models::Realtime::RealtimeSessionCreateResponse::tracing?,
-          truncation: OpenAI::Models::Realtime::realtime_truncation,
-          type: OpenAI::Models::Realtime::RealtimeSessionCreateResponse::type_
+          truncation: OpenAI::Models::Realtime::realtime_truncation
         }
 
         type audio =
@@ -446,7 +438,7 @@ module OpenAI
         end
 
         type tool =
-          OpenAI::Realtime::Models
+          OpenAI::Realtime::RealtimeFunctionTool
           | OpenAI::Realtime::RealtimeSessionCreateResponse::Tool::McpTool
 
         module Tool
@@ -712,16 +704,6 @@ module OpenAI
 
           def self?.variants: -> ::Array[OpenAI::Models::Realtime::RealtimeSessionCreateResponse::tracing]
         end
-
-        type type_ = :realtime
-
-        module Type
-          extend OpenAI::Internal::Type::Enum
-
-          REALTIME: :realtime
-
-          def self?.values: -> ::Array[OpenAI::Models::Realtime::RealtimeSessionCreateResponse::type_]
-        end
       end
     end
   end

data/sig/openai/models/realtime/realtime_tools_config_union.rbs

@@ -2,7 +2,7 @@ module OpenAI
   module Models
     module Realtime
       type realtime_tools_config_union =
-        OpenAI::Realtime::Models
+        OpenAI::Realtime::RealtimeFunctionTool
         | OpenAI::Realtime::RealtimeToolsConfigUnion::Mcp
 
       module RealtimeToolsConfigUnion