openai 0.31.0 → 0.32.0

data/lib/openai/models/vector_store_create_params.rb

@@ -14,6 +14,13 @@ module OpenAI
       #   @return [OpenAI::Models::AutoFileChunkingStrategyParam, OpenAI::Models::StaticFileChunkingStrategyObjectParam, nil]
       optional :chunking_strategy, union: -> { OpenAI::FileChunkingStrategyParam }
 
+      # @!attribute description
+      #   A description for the vector store. Can be used to describe the vector store's
+      #   purpose.
+      #
+      #   @return [String, nil]
+      optional :description, String
+
       # @!attribute expires_after
       #   The expiration policy for a vector store.
       #
@@ -45,12 +52,14 @@ module OpenAI
       #   @return [String, nil]
       optional :name, String
 
-      # @!method initialize(chunking_strategy: nil, expires_after: nil, file_ids: nil, metadata: nil, name: nil, request_options: {})
+      # @!method initialize(chunking_strategy: nil, description: nil, expires_after: nil, file_ids: nil, metadata: nil, name: nil, request_options: {})
       #   Some parameter documentations has been truncated, see
       #   {OpenAI::Models::VectorStoreCreateParams} for more details.
       #
       #   @param chunking_strategy [OpenAI::Models::AutoFileChunkingStrategyParam, OpenAI::Models::StaticFileChunkingStrategyObjectParam] The chunking strategy used to chunk the file(s). If not set, will use the `auto`
       #
+      #   @param description [String] A description for the vector store. Can be used to describe the vector store's p
+      #
       #   @param expires_after [OpenAI::Models::VectorStoreCreateParams::ExpiresAfter] The expiration policy for a vector store.
       #
       #   @param file_ids [Array<String>] A list of [File](https://platform.openai.com/docs/api-reference/files) IDs that

data/lib/openai/resources/audio/transcriptions.rb

@@ -12,7 +12,7 @@ module OpenAI
         #
         # Transcribes audio into the input language.
         #
-        # @overload create(file:, model:, chunking_strategy: nil, include: nil, language: nil, prompt: nil, response_format: nil, temperature: nil, timestamp_granularities: nil, request_options: {})
+        # @overload create(file:, model:, chunking_strategy: nil, include: nil, known_speaker_names: nil, known_speaker_references: nil, language: nil, prompt: nil, response_format: nil, temperature: nil, timestamp_granularities: nil, request_options: {})
         #
         # @param file [Pathname, StringIO, IO, String, OpenAI::FilePart] The audio file object (not file name) to transcribe, in one of these formats: fl
         #
@@ -22,6 +22,10 @@ module OpenAI
         #
         # @param include [Array<Symbol, OpenAI::Models::Audio::TranscriptionInclude>] Additional information to include in the transcription response.
         #
+        # @param known_speaker_names [Array<String>] Optional list of speaker names that correspond to the audio samples provided in
+        #
+        # @param known_speaker_references [Array<String>] Optional list of audio samples (as [data URLs](https://developer.mozilla.org/en-
+        #
         # @param language [String] The language of the input audio. Supplying the input language in [ISO-639-1](htt
         #
         # @param prompt [String] An optional text to guide the model's style or continue a previous audio segment
@@ -34,7 +38,7 @@ module OpenAI
         #
         # @param request_options [OpenAI::RequestOptions, Hash{Symbol=>Object}, nil]
         #
-        # @return [OpenAI::Models::Audio::Transcription, OpenAI::Models::Audio::TranscriptionVerbose]
+        # @return [OpenAI::Models::Audio::Transcription, OpenAI::Models::Audio::TranscriptionDiarized, OpenAI::Models::Audio::TranscriptionVerbose]
         #
         # @see OpenAI::Models::Audio::TranscriptionCreateParams
         def create(params)
@@ -61,7 +65,7 @@ module OpenAI
         #
         # Transcribes audio into the input language.
         #
-        # @overload create_streaming(file:, model:, chunking_strategy: nil, include: nil, language: nil, prompt: nil, response_format: nil, temperature: nil, timestamp_granularities: nil, request_options: {})
+        # @overload create_streaming(file:, model:, chunking_strategy: nil, include: nil, known_speaker_names: nil, known_speaker_references: nil, language: nil, prompt: nil, response_format: nil, temperature: nil, timestamp_granularities: nil, request_options: {})
         #
         # @param file [Pathname, StringIO, IO, String, OpenAI::FilePart] The audio file object (not file name) to transcribe, in one of these formats: fl
         #
@@ -71,6 +75,10 @@ module OpenAI
         #
         # @param include [Array<Symbol, OpenAI::Models::Audio::TranscriptionInclude>] Additional information to include in the transcription response.
         #
+        # @param known_speaker_names [Array<String>] Optional list of speaker names that correspond to the audio samples provided in
+        #
+        # @param known_speaker_references [Array<String>] Optional list of audio samples (as [data URLs](https://developer.mozilla.org/en-
+        #
         # @param language [String] The language of the input audio. Supplying the input language in [ISO-639-1](htt
         #
         # @param prompt [String] An optional text to guide the model's style or continue a previous audio segment
@@ -83,7 +91,7 @@ module OpenAI
         #
         # @param request_options [OpenAI::RequestOptions, Hash{Symbol=>Object}, nil]
         #
-        # @return [OpenAI::Internal::Stream<OpenAI::Models::Audio::TranscriptionTextDeltaEvent, OpenAI::Models::Audio::TranscriptionTextDoneEvent>]
+        # @return [OpenAI::Internal::Stream<OpenAI::Models::Audio::TranscriptionTextSegmentEvent, OpenAI::Models::Audio::TranscriptionTextDeltaEvent, OpenAI::Models::Audio::TranscriptionTextDoneEvent>]
         #
         # @see OpenAI::Models::Audio::TranscriptionCreateParams
         def create_streaming(params)

data/lib/openai/resources/vector_stores.rb

@@ -14,10 +14,12 @@ module OpenAI
       #
       # Create a vector store.
       #
-      # @overload create(chunking_strategy: nil, expires_after: nil, file_ids: nil, metadata: nil, name: nil, request_options: {})
+      # @overload create(chunking_strategy: nil, description: nil, expires_after: nil, file_ids: nil, metadata: nil, name: nil, request_options: {})
       #
       # @param chunking_strategy [OpenAI::Models::AutoFileChunkingStrategyParam, OpenAI::Models::StaticFileChunkingStrategyObjectParam] The chunking strategy used to chunk the file(s). If not set, will use the `auto`
       #
+      # @param description [String] A description for the vector store. Can be used to describe the vector store's p
+      #
       # @param expires_after [OpenAI::Models::VectorStoreCreateParams::ExpiresAfter] The expiration policy for a vector store.
       #
       # @param file_ids [Array<String>] A list of [File](https://platform.openai.com/docs/api-reference/files) IDs that

data/lib/openai/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module OpenAI
-  VERSION = "0.31.0"
+  VERSION = "0.32.0"
 end

data/lib/openai.rb

@@ -79,11 +79,14 @@ require_relative "openai/models/audio/speech_model"
 require_relative "openai/models/audio/transcription"
 require_relative "openai/models/audio/transcription_create_params"
 require_relative "openai/models/audio/transcription_create_response"
+require_relative "openai/models/audio/transcription_diarized"
+require_relative "openai/models/audio/transcription_diarized_segment"
 require_relative "openai/models/audio/transcription_include"
 require_relative "openai/models/audio/transcription_segment"
 require_relative "openai/models/audio/transcription_stream_event"
 require_relative "openai/models/audio/transcription_text_delta_event"
 require_relative "openai/models/audio/transcription_text_done_event"
+require_relative "openai/models/audio/transcription_text_segment_event"
 require_relative "openai/models/audio/transcription_verbose"
 require_relative "openai/models/audio/transcription_word"
 require_relative "openai/models/audio/translation"

data/rbi/openai/models/audio/transcription_create_params.rbi

@@ -21,8 +21,8 @@ module OpenAI
         attr_accessor :file
 
         # 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).
+        # `gpt-4o-mini-transcribe`, `whisper-1` (which is powered by our open source
+        # Whisper V2 model), and `gpt-4o-transcribe-diarize`.
         sig { returns(T.any(String, OpenAI::AudioModel::OrSymbol)) }
         attr_accessor :model
 
@@ -30,6 +30,8 @@ module OpenAI
         # first normalizes loudness and then uses voice activity detection (VAD) to choose
         # boundaries. `server_vad` object can be provided to tweak VAD detection
         # parameters manually. If unset, the audio is transcribed as a single block.
+        # Required when using `gpt-4o-transcribe-diarize` for inputs longer than 30
+        # seconds.
         sig do
           returns(
             T.nilable(
@@ -46,7 +48,8 @@ module OpenAI
         # return the log probabilities of the tokens in the response to understand the
         # model's confidence in the transcription. `logprobs` only works with
         # response_format set to `json` and only with the models `gpt-4o-transcribe` and
-        # `gpt-4o-mini-transcribe`.
+        # `gpt-4o-mini-transcribe`. This field is not supported when using
+        # `gpt-4o-transcribe-diarize`.
         sig do
           returns(
             T.nilable(T::Array[OpenAI::Audio::TranscriptionInclude::OrSymbol])
@@ -61,6 +64,26 @@ module OpenAI
         end
         attr_writer :include
 
+        # Optional list of speaker names that correspond to the audio samples provided in
+        # `known_speaker_references[]`. Each entry should be a short identifier (for
+        # example `customer` or `agent`). Up to 4 speakers are supported.
+        sig { returns(T.nilable(T::Array[String])) }
+        attr_reader :known_speaker_names
+
+        sig { params(known_speaker_names: T::Array[String]).void }
+        attr_writer :known_speaker_names
+
+        # Optional list of audio samples (as
+        # [data URLs](https://developer.mozilla.org/en-US/docs/Web/HTTP/Basics_of_HTTP/Data_URLs))
+        # that contain known speaker references matching `known_speaker_names[]`. Each
+        # sample must be between 2 and 10 seconds, and can use any of the same input audio
+        # formats supported by `file`.
+        sig { returns(T.nilable(T::Array[String])) }
+        attr_reader :known_speaker_references
+
+        sig { params(known_speaker_references: T::Array[String]).void }
+        attr_writer :known_speaker_references
+
         # 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.
@@ -73,7 +96,8 @@ module OpenAI
         # An optional text to guide the model's style or continue a previous audio
         # segment. The
         # [prompt](https://platform.openai.com/docs/guides/speech-to-text#prompting)
-        # should match the audio language.
+        # should match the audio language. This field is not supported when using
+        # `gpt-4o-transcribe-diarize`.
         sig { returns(T.nilable(String)) }
         attr_reader :prompt
 
@@ -81,8 +105,10 @@ module OpenAI
         attr_writer :prompt
 
         # The format of the output, in one of these options: `json`, `text`, `srt`,
-        # `verbose_json`, or `vtt`. For `gpt-4o-transcribe` and `gpt-4o-mini-transcribe`,
-        # the only supported format is `json`.
+        # `verbose_json`, `vtt`, or `diarized_json`. For `gpt-4o-transcribe` and
+        # `gpt-4o-mini-transcribe`, the only supported format is `json`. For
+        # `gpt-4o-transcribe-diarize`, the supported formats are `json`, `text`, and
+        # `diarized_json`, with `diarized_json` required to receive speaker annotations.
         sig { returns(T.nilable(OpenAI::AudioResponseFormat::OrSymbol)) }
         attr_reader :response_format
 
@@ -106,7 +132,8 @@ module OpenAI
         # `response_format` must be set `verbose_json` to use timestamp granularities.
         # Either or both of these options are supported: `word`, or `segment`. Note: There
         # is no additional latency for segment timestamps, but generating word timestamps
-        # incurs additional latency.
+        # incurs additional latency. This option is not available for
+        # `gpt-4o-transcribe-diarize`.
         sig do
           returns(
             T.nilable(
@@ -140,6 +167,8 @@ module OpenAI
                 )
               ),
             include: T::Array[OpenAI::Audio::TranscriptionInclude::OrSymbol],
+            known_speaker_names: T::Array[String],
+            known_speaker_references: T::Array[String],
             language: String,
             prompt: String,
             response_format: OpenAI::AudioResponseFormat::OrSymbol,
@@ -156,20 +185,33 @@ module OpenAI
           # flac, mp3, mp4, mpeg, mpga, m4a, ogg, wav, or webm.
           file:,
           # 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).
+          # `gpt-4o-mini-transcribe`, `whisper-1` (which is powered by our open source
+          # Whisper V2 model), and `gpt-4o-transcribe-diarize`.
           model:,
           # Controls how the audio is cut into chunks. When set to `"auto"`, the server
           # first normalizes loudness and then uses voice activity detection (VAD) to choose
           # boundaries. `server_vad` object can be provided to tweak VAD detection
           # parameters manually. If unset, the audio is transcribed as a single block.
+          # Required when using `gpt-4o-transcribe-diarize` for inputs longer than 30
+          # seconds.
           chunking_strategy: nil,
           # Additional information to include in the transcription response. `logprobs` will
           # return the log probabilities of the tokens in the response to understand the
           # model's confidence in the transcription. `logprobs` only works with
           # response_format set to `json` and only with the models `gpt-4o-transcribe` and
-          # `gpt-4o-mini-transcribe`.
+          # `gpt-4o-mini-transcribe`. This field is not supported when using
+          # `gpt-4o-transcribe-diarize`.
           include: nil,
+          # Optional list of speaker names that correspond to the audio samples provided in
+          # `known_speaker_references[]`. Each entry should be a short identifier (for
+          # example `customer` or `agent`). Up to 4 speakers are supported.
+          known_speaker_names: nil,
+          # Optional list of audio samples (as
+          # [data URLs](https://developer.mozilla.org/en-US/docs/Web/HTTP/Basics_of_HTTP/Data_URLs))
+          # that contain known speaker references matching `known_speaker_names[]`. Each
+          # sample must be between 2 and 10 seconds, and can use any of the same input audio
+          # formats supported by `file`.
+          known_speaker_references: nil,
           # 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.
@@ -177,11 +219,14 @@ module OpenAI
           # An optional text to guide the model's style or continue a previous audio
           # segment. The
           # [prompt](https://platform.openai.com/docs/guides/speech-to-text#prompting)
-          # should match the audio language.
+          # should match the audio language. This field is not supported when using
+          # `gpt-4o-transcribe-diarize`.
           prompt: nil,
           # The format of the output, in one of these options: `json`, `text`, `srt`,
-          # `verbose_json`, or `vtt`. For `gpt-4o-transcribe` and `gpt-4o-mini-transcribe`,
-          # the only supported format is `json`.
+          # `verbose_json`, `vtt`, or `diarized_json`. For `gpt-4o-transcribe` and
+          # `gpt-4o-mini-transcribe`, the only supported format is `json`. For
+          # `gpt-4o-transcribe-diarize`, the supported formats are `json`, `text`, and
+          # `diarized_json`, with `diarized_json` required to receive speaker annotations.
           response_format: nil,
           # The sampling temperature, between 0 and 1. Higher values like 0.8 will make the
           # output more random, while lower values like 0.2 will make it more focused and
@@ -193,7 +238,8 @@ module OpenAI
           # `response_format` must be set `verbose_json` to use timestamp granularities.
           # Either or both of these options are supported: `word`, or `segment`. Note: There
           # is no additional latency for segment timestamps, but generating word timestamps
-          # incurs additional latency.
+          # incurs additional latency. This option is not available for
+          # `gpt-4o-transcribe-diarize`.
           timestamp_granularities: nil,
           request_options: {}
         )
@@ -212,6 +258,8 @@ module OpenAI
                   )
                 ),
               include: T::Array[OpenAI::Audio::TranscriptionInclude::OrSymbol],
+              known_speaker_names: T::Array[String],
+              known_speaker_references: T::Array[String],
               language: String,
               prompt: String,
               response_format: OpenAI::AudioResponseFormat::OrSymbol,
@@ -228,8 +276,8 @@ module OpenAI
         end
 
         # 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).
+        # `gpt-4o-mini-transcribe`, `whisper-1` (which is powered by our open source
+        # Whisper V2 model), and `gpt-4o-transcribe-diarize`.
         module Model
           extend OpenAI::Internal::Type::Union
 
@@ -251,6 +299,8 @@ module OpenAI
         # first normalizes loudness and then uses voice activity detection (VAD) to choose
         # boundaries. `server_vad` object can be provided to tweak VAD detection
         # parameters manually. If unset, the audio is transcribed as a single block.
+        # Required when using `gpt-4o-transcribe-diarize` for inputs longer than 30
+        # seconds.
         module ChunkingStrategy
           extend OpenAI::Internal::Type::Union
 

data/rbi/openai/models/audio/transcription_create_response.rbi

@@ -12,6 +12,7 @@ module OpenAI
           T.type_alias do
             T.any(
               OpenAI::Audio::Transcription,
+              OpenAI::Audio::TranscriptionDiarized,
               OpenAI::Audio::TranscriptionVerbose
             )
           end

data/rbi/openai/models/audio/transcription_diarized.rbi

@@ -0,0 +1,281 @@
+# typed: strong
+
+module OpenAI
+  module Models
+    module Audio
+      class TranscriptionDiarized < OpenAI::Internal::Type::BaseModel
+        OrHash =
+          T.type_alias do
+            T.any(
+              OpenAI::Audio::TranscriptionDiarized,
+              OpenAI::Internal::AnyHash
+            )
+          end
+
+        # Duration of the input audio in seconds.
+        sig { returns(Float) }
+        attr_accessor :duration
+
+        # Segments of the transcript annotated with timestamps and speaker labels.
+        sig { returns(T::Array[OpenAI::Audio::TranscriptionDiarizedSegment]) }
+        attr_accessor :segments
+
+        # The type of task that was run. Always `transcribe`.
+        sig { returns(Symbol) }
+        attr_accessor :task
+
+        # The concatenated transcript text for the entire audio input.
+        sig { returns(String) }
+        attr_accessor :text
+
+        # Token or duration usage statistics for the request.
+        sig do
+          returns(
+            T.nilable(OpenAI::Audio::TranscriptionDiarized::Usage::Variants)
+          )
+        end
+        attr_reader :usage
+
+        sig do
+          params(
+            usage:
+              T.any(
+                OpenAI::Audio::TranscriptionDiarized::Usage::Tokens::OrHash,
+                OpenAI::Audio::TranscriptionDiarized::Usage::Duration::OrHash
+              )
+          ).void
+        end
+        attr_writer :usage
+
+        # Represents a diarized transcription response returned by the model, including
+        # the combined transcript and speaker-segment annotations.
+        sig do
+          params(
+            duration: Float,
+            segments:
+              T::Array[OpenAI::Audio::TranscriptionDiarizedSegment::OrHash],
+            text: String,
+            usage:
+              T.any(
+                OpenAI::Audio::TranscriptionDiarized::Usage::Tokens::OrHash,
+                OpenAI::Audio::TranscriptionDiarized::Usage::Duration::OrHash
+              ),
+            task: Symbol
+          ).returns(T.attached_class)
+        end
+        def self.new(
+          # Duration of the input audio in seconds.
+          duration:,
+          # Segments of the transcript annotated with timestamps and speaker labels.
+          segments:,
+          # The concatenated transcript text for the entire audio input.
+          text:,
+          # Token or duration usage statistics for the request.
+          usage: nil,
+          # The type of task that was run. Always `transcribe`.
+          task: :transcribe
+        )
+        end
+
+        sig do
+          override.returns(
+            {
+              duration: Float,
+              segments: T::Array[OpenAI::Audio::TranscriptionDiarizedSegment],
+              task: Symbol,
+              text: String,
+              usage: OpenAI::Audio::TranscriptionDiarized::Usage::Variants
+            }
+          )
+        end
+        def to_hash
+        end
+
+        # Token or duration usage statistics for the request.
+        module Usage
+          extend OpenAI::Internal::Type::Union
+
+          Variants =
+            T.type_alias do
+              T.any(
+                OpenAI::Audio::TranscriptionDiarized::Usage::Tokens,
+                OpenAI::Audio::TranscriptionDiarized::Usage::Duration
+              )
+            end
+
+          class Tokens < OpenAI::Internal::Type::BaseModel
+            OrHash =
+              T.type_alias do
+                T.any(
+                  OpenAI::Audio::TranscriptionDiarized::Usage::Tokens,
+                  OpenAI::Internal::AnyHash
+                )
+              end
+
+            # Number of input tokens billed for this request.
+            sig { returns(Integer) }
+            attr_accessor :input_tokens
+
+            # Number of output tokens generated.
+            sig { returns(Integer) }
+            attr_accessor :output_tokens
+
+            # Total number of tokens used (input + output).
+            sig { returns(Integer) }
+            attr_accessor :total_tokens
+
+            # The type of the usage object. Always `tokens` for this variant.
+            sig { returns(Symbol) }
+            attr_accessor :type
+
+            # Details about the input tokens billed for this request.
+            sig do
+              returns(
+                T.nilable(
+                  OpenAI::Audio::TranscriptionDiarized::Usage::Tokens::InputTokenDetails
+                )
+              )
+            end
+            attr_reader :input_token_details
+
+            sig do
+              params(
+                input_token_details:
+                  OpenAI::Audio::TranscriptionDiarized::Usage::Tokens::InputTokenDetails::OrHash
+              ).void
+            end
+            attr_writer :input_token_details
+
+            # Usage statistics for models billed by token usage.
+            sig do
+              params(
+                input_tokens: Integer,
+                output_tokens: Integer,
+                total_tokens: Integer,
+                input_token_details:
+                  OpenAI::Audio::TranscriptionDiarized::Usage::Tokens::InputTokenDetails::OrHash,
+                type: Symbol
+              ).returns(T.attached_class)
+            end
+            def self.new(
+              # Number of input tokens billed for this request.
+              input_tokens:,
+              # Number of output tokens generated.
+              output_tokens:,
+              # Total number of tokens used (input + output).
+              total_tokens:,
+              # Details about the input tokens billed for this request.
+              input_token_details: nil,
+              # The type of the usage object. Always `tokens` for this variant.
+              type: :tokens
+            )
+            end
+
+            sig do
+              override.returns(
+                {
+                  input_tokens: Integer,
+                  output_tokens: Integer,
+                  total_tokens: Integer,
+                  type: Symbol,
+                  input_token_details:
+                    OpenAI::Audio::TranscriptionDiarized::Usage::Tokens::InputTokenDetails
+                }
+              )
+            end
+            def to_hash
+            end
+
+            class InputTokenDetails < OpenAI::Internal::Type::BaseModel
+              OrHash =
+                T.type_alias do
+                  T.any(
+                    OpenAI::Audio::TranscriptionDiarized::Usage::Tokens::InputTokenDetails,
+                    OpenAI::Internal::AnyHash
+                  )
+                end
+
+              # Number of audio tokens billed for this request.
+              sig { returns(T.nilable(Integer)) }
+              attr_reader :audio_tokens
+
+              sig { params(audio_tokens: Integer).void }
+              attr_writer :audio_tokens
+
+              # Number of text tokens billed for this request.
+              sig { returns(T.nilable(Integer)) }
+              attr_reader :text_tokens
+
+              sig { params(text_tokens: Integer).void }
+              attr_writer :text_tokens
+
+              # Details about the input tokens billed for this request.
+              sig do
+                params(audio_tokens: Integer, text_tokens: Integer).returns(
+                  T.attached_class
+                )
+              end
+              def self.new(
+                # Number of audio tokens billed for this request.
+                audio_tokens: nil,
+                # Number of text tokens billed for this request.
+                text_tokens: nil
+              )
+              end
+
+              sig do
+                override.returns(
+                  { audio_tokens: Integer, text_tokens: Integer }
+                )
+              end
+              def to_hash
+              end
+            end
+          end
+
+          class Duration < OpenAI::Internal::Type::BaseModel
+            OrHash =
+              T.type_alias do
+                T.any(
+                  OpenAI::Audio::TranscriptionDiarized::Usage::Duration,
+                  OpenAI::Internal::AnyHash
+                )
+              end
+
+            # Duration of the input audio in seconds.
+            sig { returns(Float) }
+            attr_accessor :seconds
+
+            # The type of the usage object. Always `duration` for this variant.
+            sig { returns(Symbol) }
+            attr_accessor :type
+
+            # Usage statistics for models billed by audio input duration.
+            sig do
+              params(seconds: Float, type: Symbol).returns(T.attached_class)
+            end
+            def self.new(
+              # Duration of the input audio in seconds.
+              seconds:,
+              # The type of the usage object. Always `duration` for this variant.
+              type: :duration
+            )
+            end
+
+            sig { override.returns({ seconds: Float, type: Symbol }) }
+            def to_hash
+            end
+          end
+
+          sig do
+            override.returns(
+              T::Array[OpenAI::Audio::TranscriptionDiarized::Usage::Variants]
+            )
+          end
+          def self.variants
+          end
+        end
+      end
+    end
+  end
+end