openai 0.22.1 → 0.23.0

data/lib/openai/models/realtime/realtime_transcription_session_create_request.rb

@@ -4,14 +4,6 @@ module OpenAI
   module Models
     module Realtime
       class RealtimeTranscriptionSessionCreateRequest < OpenAI::Internal::Type::BaseModel
-        # @!attribute model
-        #   ID of the model to use. The options are `gpt-4o-transcribe`,
-        #   `gpt-4o-mini-transcribe`, and `whisper-1` (which is powered by our open source
-        #   Whisper V2 model).
-        #
-        #   @return [String, Symbol, OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::Model]
-        required :model, union: -> { OpenAI::Realtime::RealtimeTranscriptionSessionCreateRequest::Model }
-
         # @!attribute type
         #   The type of session to create. Always `transcription` for transcription
         #   sessions.
@@ -19,106 +11,35 @@ module OpenAI
         #   @return [Symbol, :transcription]
         required :type, const: :transcription
 
+        # @!attribute audio
+        #   Configuration for input and output audio.
+        #
+        #   @return [OpenAI::Models::Realtime::RealtimeTranscriptionSessionAudio, nil]
+        optional :audio, -> { OpenAI::Realtime::RealtimeTranscriptionSessionAudio }
+
         # @!attribute include
-        #   The set of items to include in the transcription. Current available items are:
+        #   Additional fields to include in server outputs.
         #
-        #   - `item.input_audio_transcription.logprobs`
+        #   `item.input_audio_transcription.logprobs`: Include logprobs for input audio
+        #   transcription.
         #
         #   @return [Array<Symbol, OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::Include>, nil]
         optional :include,
                  -> { OpenAI::Internal::Type::ArrayOf[enum: OpenAI::Realtime::RealtimeTranscriptionSessionCreateRequest::Include] }
 
-        # @!attribute input_audio_format
-        #   The format of input audio. Options are `pcm16`, `g711_ulaw`, or `g711_alaw`. For
-        #   `pcm16`, input audio must be 16-bit PCM at a 24kHz sample rate, single channel
-        #   (mono), and little-endian byte order.
-        #
-        #   @return [Symbol, OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::InputAudioFormat, nil]
-        optional :input_audio_format,
-                 enum: -> { OpenAI::Realtime::RealtimeTranscriptionSessionCreateRequest::InputAudioFormat }
-
-        # @!attribute input_audio_noise_reduction
-        #   Configuration for input audio noise reduction. This can be set to `null` to turn
-        #   off. Noise reduction filters audio added to the input audio buffer before it is
-        #   sent to VAD and the model. Filtering the audio can improve VAD and turn
-        #   detection accuracy (reducing false positives) and model performance by improving
-        #   perception of the input audio.
-        #
-        #   @return [OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::InputAudioNoiseReduction, nil]
-        optional :input_audio_noise_reduction,
-                 -> { OpenAI::Realtime::RealtimeTranscriptionSessionCreateRequest::InputAudioNoiseReduction }
-
-        # @!attribute input_audio_transcription
-        #   Configuration for input audio transcription. The client can optionally set the
-        #   language and prompt for transcription, these offer additional guidance to the
-        #   transcription service.
-        #
-        #   @return [OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::InputAudioTranscription, nil]
-        optional :input_audio_transcription,
-                 -> { OpenAI::Realtime::RealtimeTranscriptionSessionCreateRequest::InputAudioTranscription }
-
-        # @!attribute turn_detection
-        #   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.
-        #
-        #   @return [OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::TurnDetection, nil]
-        optional :turn_detection,
-                 -> { OpenAI::Realtime::RealtimeTranscriptionSessionCreateRequest::TurnDetection }
-
-        # @!method initialize(model:, include: nil, input_audio_format: nil, input_audio_noise_reduction: nil, input_audio_transcription: nil, turn_detection: nil, type: :transcription)
+        # @!method initialize(audio: nil, include: nil, type: :transcription)
         #   Some parameter documentations has been truncated, see
         #   {OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest} for more
         #   details.
         #
         #   Realtime transcription session object configuration.
         #
-        #   @param model [String, Symbol, OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::Model] ID of the model to use. The options are `gpt-4o-transcribe`, `gpt-4o-mini-transc
-        #
-        #   @param include [Array<Symbol, OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::Include>] The set of items to include in the transcription. Current available items are:
-        #
-        #   @param input_audio_format [Symbol, OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::InputAudioFormat] The format of input audio. Options are `pcm16`, `g711_ulaw`, or `g711_alaw`.
-        #
-        #   @param input_audio_noise_reduction [OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::InputAudioNoiseReduction] Configuration for input audio noise reduction. This can be set to `null` to turn
+        #   @param audio [OpenAI::Models::Realtime::RealtimeTranscriptionSessionAudio] Configuration for input and output audio.
         #
-        #   @param input_audio_transcription [OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::InputAudioTranscription] Configuration for input audio transcription. The client can optionally set the l
-        #
-        #   @param turn_detection [OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::TurnDetection] Configuration for turn detection. Can be set to `null` to turn off. Server VAD m
+        #   @param include [Array<Symbol, OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::Include>] Additional fields to include in server outputs.
         #
         #   @param type [Symbol, :transcription] The type of session to create. Always `transcription` for transcription sessions
 
-        # ID of the model to use. The options are `gpt-4o-transcribe`,
-        # `gpt-4o-mini-transcribe`, and `whisper-1` (which is powered by our open source
-        # Whisper V2 model).
-        #
-        # @see OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest#model
-        module Model
-          extend OpenAI::Internal::Type::Union
-
-          variant String
-
-          variant const: -> { OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::Model::WHISPER_1 }
-
-          variant const: -> { OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::Model::GPT_4O_TRANSCRIBE }
-
-          variant const: -> { OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::Model::GPT_4O_MINI_TRANSCRIBE }
-
-          # @!method self.variants
-          #   @return [Array(String, Symbol)]
-
-          define_sorbet_constant!(:Variants) do
-            T.type_alias { T.any(String, OpenAI::Realtime::RealtimeTranscriptionSessionCreateRequest::Model::TaggedSymbol) }
-          end
-
-          # @!group
-
-          WHISPER_1 = :"whisper-1"
-          GPT_4O_TRANSCRIBE = :"gpt-4o-transcribe"
-          GPT_4O_MINI_TRANSCRIBE = :"gpt-4o-mini-transcribe"
-
-          # @!endgroup
-        end
-
         module Include
           extend OpenAI::Internal::Type::Enum
 
@@ -127,185 +48,6 @@ module OpenAI
           # @!method self.values
           #   @return [Array<Symbol>]
         end
-
-        # The format of input audio. Options are `pcm16`, `g711_ulaw`, or `g711_alaw`. For
-        # `pcm16`, input audio must be 16-bit PCM at a 24kHz sample rate, single channel
-        # (mono), and little-endian byte order.
-        #
-        # @see OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest#input_audio_format
-        module InputAudioFormat
-          extend OpenAI::Internal::Type::Enum
-
-          PCM16 = :pcm16
-          G711_ULAW = :g711_ulaw
-          G711_ALAW = :g711_alaw
-
-          # @!method self.values
-          #   @return [Array<Symbol>]
-        end
-
-        # @see OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest#input_audio_noise_reduction
-        class InputAudioNoiseReduction < OpenAI::Internal::Type::BaseModel
-          # @!attribute type
-          #   Type of noise reduction. `near_field` is for close-talking microphones such as
-          #   headphones, `far_field` is for far-field microphones such as laptop or
-          #   conference room microphones.
-          #
-          #   @return [Symbol, OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::InputAudioNoiseReduction::Type, nil]
-          optional :type,
-                   enum: -> { OpenAI::Realtime::RealtimeTranscriptionSessionCreateRequest::InputAudioNoiseReduction::Type }
-
-          # @!method initialize(type: nil)
-          #   Some parameter documentations has been truncated, see
-          #   {OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::InputAudioNoiseReduction}
-          #   for more details.
-          #
-          #   Configuration for input audio noise reduction. This can be set to `null` to turn
-          #   off. Noise reduction filters audio added to the input audio buffer before it is
-          #   sent to VAD and the model. Filtering the audio can improve VAD and turn
-          #   detection accuracy (reducing false positives) and model performance by improving
-          #   perception of the input audio.
-          #
-          #   @param type [Symbol, OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::InputAudioNoiseReduction::Type] Type of noise reduction. `near_field` is for close-talking microphones such as h
-
-          # Type of noise reduction. `near_field` is for close-talking microphones such as
-          # headphones, `far_field` is for far-field microphones such as laptop or
-          # conference room microphones.
-          #
-          # @see OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::InputAudioNoiseReduction#type
-          module Type
-            extend OpenAI::Internal::Type::Enum
-
-            NEAR_FIELD = :near_field
-            FAR_FIELD = :far_field
-
-            # @!method self.values
-            #   @return [Array<Symbol>]
-          end
-        end
-
-        # @see OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest#input_audio_transcription
-        class InputAudioTranscription < OpenAI::Internal::Type::BaseModel
-          # @!attribute language
-          #   The language of the input audio. Supplying the input language in
-          #   [ISO-639-1](https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes) (e.g. `en`)
-          #   format will improve accuracy and latency.
-          #
-          #   @return [String, nil]
-          optional :language, String
-
-          # @!attribute model
-          #   The model to use for transcription, current options are `gpt-4o-transcribe`,
-          #   `gpt-4o-mini-transcribe`, and `whisper-1`.
-          #
-          #   @return [Symbol, OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::InputAudioTranscription::Model, nil]
-          optional :model,
-                   enum: -> { OpenAI::Realtime::RealtimeTranscriptionSessionCreateRequest::InputAudioTranscription::Model }
-
-          # @!attribute prompt
-          #   An optional text to guide the model's style or continue a previous audio
-          #   segment. For `whisper-1`, the
-          #   [prompt is a list of keywords](https://platform.openai.com/docs/guides/speech-to-text#prompting).
-          #   For `gpt-4o-transcribe` models, the prompt is a free text string, for example
-          #   "expect words related to technology".
-          #
-          #   @return [String, nil]
-          optional :prompt, String
-
-          # @!method initialize(language: nil, model: nil, prompt: nil)
-          #   Some parameter documentations has been truncated, see
-          #   {OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::InputAudioTranscription}
-          #   for more details.
-          #
-          #   Configuration for input audio transcription. The client can optionally set the
-          #   language and prompt for transcription, these offer additional guidance to the
-          #   transcription service.
-          #
-          #   @param language [String] The language of the input audio. Supplying the input language in
-          #
-          #   @param model [Symbol, OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::InputAudioTranscription::Model] The model to use for transcription, current options are `gpt-4o-transcribe`, `gp
-          #
-          #   @param prompt [String] An optional text to guide the model's style or continue a previous audio
-
-          # The model to use for transcription, current options are `gpt-4o-transcribe`,
-          # `gpt-4o-mini-transcribe`, and `whisper-1`.
-          #
-          # @see OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::InputAudioTranscription#model
-          module Model
-            extend OpenAI::Internal::Type::Enum
-
-            GPT_4O_TRANSCRIBE = :"gpt-4o-transcribe"
-            GPT_4O_MINI_TRANSCRIBE = :"gpt-4o-mini-transcribe"
-            WHISPER_1 = :"whisper-1"
-
-            # @!method self.values
-            #   @return [Array<Symbol>]
-          end
-        end
-
-        # @see OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest#turn_detection
-        class TurnDetection < OpenAI::Internal::Type::BaseModel
-          # @!attribute prefix_padding_ms
-          #   Amount of audio to include before the VAD detected speech (in milliseconds).
-          #   Defaults to 300ms.
-          #
-          #   @return [Integer, nil]
-          optional :prefix_padding_ms, Integer
-
-          # @!attribute silence_duration_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.
-          #
-          #   @return [Integer, nil]
-          optional :silence_duration_ms, Integer
-
-          # @!attribute threshold
-          #   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.
-          #
-          #   @return [Float, nil]
-          optional :threshold, Float
-
-          # @!attribute type
-          #   Type of turn detection. Only `server_vad` is currently supported for
-          #   transcription sessions.
-          #
-          #   @return [Symbol, OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::TurnDetection::Type, nil]
-          optional :type,
-                   enum: -> { OpenAI::Realtime::RealtimeTranscriptionSessionCreateRequest::TurnDetection::Type }
-
-          # @!method initialize(prefix_padding_ms: nil, silence_duration_ms: nil, threshold: nil, type: nil)
-          #   Some parameter documentations has been truncated, see
-          #   {OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::TurnDetection}
-          #   for more details.
-          #
-          #   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.
-          #
-          #   @param prefix_padding_ms [Integer] Amount of audio to include before the VAD detected speech (in
-          #
-          #   @param silence_duration_ms [Integer] Duration of silence to detect speech stop (in milliseconds). Defaults
-          #
-          #   @param threshold [Float] Activation threshold for VAD (0.0 to 1.0), this defaults to 0.5. A
-          #
-          #   @param type [Symbol, OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::TurnDetection::Type] Type of turn detection. Only `server_vad` is currently supported for transcripti
-
-          # Type of turn detection. Only `server_vad` is currently supported for
-          # transcription sessions.
-          #
-          # @see OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateRequest::TurnDetection#type
-          module Type
-            extend OpenAI::Internal::Type::Enum
-
-            SERVER_VAD = :server_vad
-
-            # @!method self.values
-            #   @return [Array<Symbol>]
-          end
-        end
       end
     end
   end

data/lib/openai/models/realtime/realtime_transcription_session_create_response.rb

@@ -0,0 +1,78 @@
+# frozen_string_literal: true
+
+module OpenAI
+  module Models
+    module Realtime
+      class RealtimeTranscriptionSessionCreateResponse < OpenAI::Internal::Type::BaseModel
+        # @!attribute client_secret
+        #   Ephemeral key returned by the API. Only present when the session is created on
+        #   the server via REST API.
+        #
+        #   @return [OpenAI::Models::Realtime::RealtimeTranscriptionSessionClientSecret]
+        required :client_secret, -> { OpenAI::Realtime::RealtimeTranscriptionSessionClientSecret }
+
+        # @!attribute input_audio_format
+        #   The format of input audio. Options are `pcm16`, `g711_ulaw`, or `g711_alaw`.
+        #
+        #   @return [String, nil]
+        optional :input_audio_format, String
+
+        # @!attribute input_audio_transcription
+        #   Configuration of the transcription model.
+        #
+        #   @return [OpenAI::Models::Realtime::RealtimeTranscriptionSessionInputAudioTranscription, nil]
+        optional :input_audio_transcription,
+                 -> { OpenAI::Realtime::RealtimeTranscriptionSessionInputAudioTranscription }
+
+        # @!attribute modalities
+        #   The set of modalities the model can respond with. To disable audio, set this to
+        #   ["text"].
+        #
+        #   @return [Array<Symbol, OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateResponse::Modality>, nil]
+        optional :modalities,
+                 -> { OpenAI::Internal::Type::ArrayOf[enum: OpenAI::Realtime::RealtimeTranscriptionSessionCreateResponse::Modality] }
+
+        # @!attribute turn_detection
+        #   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.
+        #
+        #   @return [OpenAI::Models::Realtime::RealtimeTranscriptionSessionTurnDetection, nil]
+        optional :turn_detection, -> { OpenAI::Realtime::RealtimeTranscriptionSessionTurnDetection }
+
+        # @!method initialize(client_secret:, input_audio_format: nil, input_audio_transcription: nil, modalities: nil, turn_detection: nil)
+        #   Some parameter documentations has been truncated, see
+        #   {OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateResponse} for more
+        #   details.
+        #
+        #   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.
+        #
+        #   @param client_secret [OpenAI::Models::Realtime::RealtimeTranscriptionSessionClientSecret] Ephemeral key returned by the API. Only present when the session is
+        #
+        #   @param input_audio_format [String] The format of input audio. Options are `pcm16`, `g711_ulaw`, or `g711_alaw`.
+        #
+        #   @param input_audio_transcription [OpenAI::Models::Realtime::RealtimeTranscriptionSessionInputAudioTranscription] Configuration of the transcription model.
+        #
+        #   @param modalities [Array<Symbol, OpenAI::Models::Realtime::RealtimeTranscriptionSessionCreateResponse::Modality>] The set of modalities the model can respond with. To disable audio,
+        #
+        #   @param turn_detection [OpenAI::Models::Realtime::RealtimeTranscriptionSessionTurnDetection] Configuration for turn detection. Can be set to `null` to turn off. Server
+
+        module Modality
+          extend OpenAI::Internal::Type::Enum
+
+          TEXT = :text
+          AUDIO = :audio
+
+          # @!method self.values
+          #   @return [Array<Symbol>]
+        end
+      end
+    end
+
+    RealtimeTranscriptionSessionCreateResponse = Realtime::RealtimeTranscriptionSessionCreateResponse
+  end
+end

data/lib/openai/models/realtime/realtime_transcription_session_input_audio_transcription.rb

@@ -0,0 +1,66 @@
+# frozen_string_literal: true
+
+module OpenAI
+  module Models
+    module Realtime
+      class RealtimeTranscriptionSessionInputAudioTranscription < OpenAI::Internal::Type::BaseModel
+        # @!attribute language
+        #   The language of the input audio. Supplying the input language in
+        #   [ISO-639-1](https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes) (e.g. `en`)
+        #   format will improve accuracy and latency.
+        #
+        #   @return [String, nil]
+        optional :language, String
+
+        # @!attribute model
+        #   The model to use for transcription. Current options are `whisper-1`,
+        #   `gpt-4o-transcribe-latest`, `gpt-4o-mini-transcribe`, and `gpt-4o-transcribe`.
+        #
+        #   @return [Symbol, OpenAI::Models::Realtime::RealtimeTranscriptionSessionInputAudioTranscription::Model, nil]
+        optional :model, enum: -> { OpenAI::Realtime::RealtimeTranscriptionSessionInputAudioTranscription::Model }
+
+        # @!attribute prompt
+        #   An optional text to guide the model's style or continue a previous audio
+        #   segment. For `whisper-1`, the
+        #   [prompt is a list of keywords](https://platform.openai.com/docs/guides/speech-to-text#prompting).
+        #   For `gpt-4o-transcribe` models, the prompt is a free text string, for example
+        #   "expect words related to technology".
+        #
+        #   @return [String, nil]
+        optional :prompt, String
+
+        # @!method initialize(language: nil, model: nil, prompt: nil)
+        #   Some parameter documentations has been truncated, see
+        #   {OpenAI::Models::Realtime::RealtimeTranscriptionSessionInputAudioTranscription}
+        #   for more details.
+        #
+        #   Configuration of the transcription model.
+        #
+        #   @param language [String] The language of the input audio. Supplying the input language in
+        #
+        #   @param model [Symbol, OpenAI::Models::Realtime::RealtimeTranscriptionSessionInputAudioTranscription::Model] The model to use for transcription. Current options are `whisper-1`, `gpt-4o-tra
+        #
+        #   @param prompt [String] An optional text to guide the model's style or continue a previous audio
+
+        # The model to use for transcription. Current options are `whisper-1`,
+        # `gpt-4o-transcribe-latest`, `gpt-4o-mini-transcribe`, and `gpt-4o-transcribe`.
+        #
+        # @see OpenAI::Models::Realtime::RealtimeTranscriptionSessionInputAudioTranscription#model
+        module Model
+          extend OpenAI::Internal::Type::Enum
+
+          WHISPER_1 = :"whisper-1"
+          GPT_4O_TRANSCRIBE_LATEST = :"gpt-4o-transcribe-latest"
+          GPT_4O_MINI_TRANSCRIBE = :"gpt-4o-mini-transcribe"
+          GPT_4O_TRANSCRIBE = :"gpt-4o-transcribe"
+
+          # @!method self.values
+          #   @return [Array<Symbol>]
+        end
+      end
+    end
+
+    RealtimeTranscriptionSessionInputAudioTranscription =
+      Realtime::RealtimeTranscriptionSessionInputAudioTranscription
+  end
+end

data/lib/openai/models/realtime/realtime_transcription_session_turn_detection.rb

@@ -0,0 +1,57 @@
+# frozen_string_literal: true
+
+module OpenAI
+  module Models
+    module Realtime
+      class RealtimeTranscriptionSessionTurnDetection < OpenAI::Internal::Type::BaseModel
+        # @!attribute prefix_padding_ms
+        #   Amount of audio to include before the VAD detected speech (in milliseconds).
+        #   Defaults to 300ms.
+        #
+        #   @return [Integer, nil]
+        optional :prefix_padding_ms, Integer
+
+        # @!attribute silence_duration_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.
+        #
+        #   @return [Integer, nil]
+        optional :silence_duration_ms, Integer
+
+        # @!attribute threshold
+        #   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.
+        #
+        #   @return [Float, nil]
+        optional :threshold, Float
+
+        # @!attribute type
+        #   Type of turn detection, only `server_vad` is currently supported.
+        #
+        #   @return [String, nil]
+        optional :type, String
+
+        # @!method initialize(prefix_padding_ms: nil, silence_duration_ms: nil, threshold: nil, type: nil)
+        #   Some parameter documentations has been truncated, see
+        #   {OpenAI::Models::Realtime::RealtimeTranscriptionSessionTurnDetection} for more
+        #   details.
+        #
+        #   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.
+        #
+        #   @param prefix_padding_ms [Integer] Amount of audio to include before the VAD detected speech (in
+        #
+        #   @param silence_duration_ms [Integer] Duration of silence to detect speech stop (in milliseconds). Defaults
+        #
+        #   @param threshold [Float] Activation threshold for VAD (0.0 to 1.0), this defaults to 0.5. A
+        #
+        #   @param type [String] Type of turn detection, only `server_vad` is currently supported.
+      end
+    end
+
+    RealtimeTranscriptionSessionTurnDetection = Realtime::RealtimeTranscriptionSessionTurnDetection
+  end
+end

data/lib/openai/models/realtime/realtime_truncation.rb

@@ -4,18 +4,19 @@ module OpenAI
   module Models
     module Realtime
       # Controls how the realtime conversation is truncated prior to model inference.
-      # The default is `auto`. When set to `retention_ratio`, the server retains a
-      # fraction of the conversation tokens prior to the instructions.
+      # The default is `auto`.
       module RealtimeTruncation
         extend OpenAI::Internal::Type::Union
 
-        # The truncation strategy to use for the session.
+        # The truncation strategy to use for the session. `auto` is the default truncation strategy. `disabled` will disable truncation and emit errors when the conversation exceeds the input token limit.
         variant enum: -> { OpenAI::Realtime::RealtimeTruncation::RealtimeTruncationStrategy }
 
-        # Retain a fraction of the conversation tokens.
-        variant -> { OpenAI::Realtime::RealtimeTruncation::RetentionRatioTruncation }
+        # Retain a fraction of the conversation tokens when the conversation exceeds the input token limit. This allows you to amortize truncations across multiple turns, which can help improve cached token usage.
+        variant -> { OpenAI::Realtime::RealtimeTruncationRetentionRatio }
 
-        # The truncation strategy to use for the session.
+        # The truncation strategy to use for the session. `auto` is the default truncation
+        # strategy. `disabled` will disable truncation and emit errors when the
+        # conversation exceeds the input token limit.
         module RealtimeTruncationStrategy
           extend OpenAI::Internal::Type::Enum
 
@@ -26,41 +27,8 @@ module OpenAI
           #   @return [Array<Symbol>]
         end
 
-        class RetentionRatioTruncation < OpenAI::Internal::Type::BaseModel
-          # @!attribute retention_ratio
-          #   Fraction of pre-instruction conversation tokens to retain (0.0 - 1.0).
-          #
-          #   @return [Float]
-          required :retention_ratio, Float
-
-          # @!attribute type
-          #   Use retention ratio truncation.
-          #
-          #   @return [Symbol, :retention_ratio]
-          required :type, const: :retention_ratio
-
-          # @!attribute post_instructions_token_limit
-          #   Optional cap on tokens allowed after the instructions.
-          #
-          #   @return [Integer, nil]
-          optional :post_instructions_token_limit, Integer, nil?: true
-
-          # @!method initialize(retention_ratio:, post_instructions_token_limit: nil, type: :retention_ratio)
-          #   Some parameter documentations has been truncated, see
-          #   {OpenAI::Models::Realtime::RealtimeTruncation::RetentionRatioTruncation} for
-          #   more details.
-          #
-          #   Retain a fraction of the conversation tokens.
-          #
-          #   @param retention_ratio [Float] Fraction of pre-instruction conversation tokens to retain (0.0 - 1.0).
-          #
-          #   @param post_instructions_token_limit [Integer, nil] Optional cap on tokens allowed after the instructions.
-          #
-          #   @param type [Symbol, :retention_ratio] Use retention ratio truncation.
-        end
-
         # @!method self.variants
-        #   @return [Array(Symbol, OpenAI::Models::Realtime::RealtimeTruncation::RealtimeTruncationStrategy, OpenAI::Models::Realtime::RealtimeTruncation::RetentionRatioTruncation)]
+        #   @return [Array(Symbol, OpenAI::Models::Realtime::RealtimeTruncation::RealtimeTruncationStrategy, OpenAI::Models::Realtime::RealtimeTruncationRetentionRatio)]
       end
     end
   end

data/lib/openai/models/realtime/realtime_truncation_retention_ratio.rb

@@ -0,0 +1,34 @@
+# frozen_string_literal: true
+
+module OpenAI
+  module Models
+    module Realtime
+      class RealtimeTruncationRetentionRatio < OpenAI::Internal::Type::BaseModel
+        # @!attribute retention_ratio
+        #   Fraction of post-instruction conversation tokens to retain (0.0 - 1.0) when the
+        #   conversation exceeds the input token limit.
+        #
+        #   @return [Float]
+        required :retention_ratio, Float
+
+        # @!attribute type
+        #   Use retention ratio truncation.
+        #
+        #   @return [Symbol, :retention_ratio]
+        required :type, const: :retention_ratio
+
+        # @!method initialize(retention_ratio:, type: :retention_ratio)
+        #   Some parameter documentations has been truncated, see
+        #   {OpenAI::Models::Realtime::RealtimeTruncationRetentionRatio} for more details.
+        #
+        #   Retain a fraction of the conversation tokens when the conversation exceeds the
+        #   input token limit. This allows you to amortize truncations across multiple
+        #   turns, which can help improve cached token usage.
+        #
+        #   @param retention_ratio [Float] Fraction of post-instruction conversation tokens to retain (0.0 - 1.0) when the
+        #
+        #   @param type [Symbol, :retention_ratio] Use retention ratio truncation.
+      end
+    end
+  end
+end

data/lib/openai/models/realtime/response_cancel_event.rb

@@ -29,7 +29,9 @@ module OpenAI
         #
         #   Send this event to cancel an in-progress response. The server will respond with
         #   a `response.done` event with a status of `response.status=cancelled`. If there
-        #   is no response to cancel, the server will respond with an error.
+        #   is no response to cancel, the server will respond with an error. It's safe to
+        #   call `response.cancel` even if no response is in progress, an error will be
+        #   returned the session will remain unaffected.
         #
         #   @param event_id [String] Optional client-generated ID used to identify this event.
         #