Added response_format option to openai_transcription processor

docs/modules/components/pages/processors/openai_transcription.adoc

@@ -59,6 +59,7 @@ openai_transcription:
   file: "" # No default (required)
   language: en # No default (optional)
   prompt: "" # No default (optional)
+  response_format: json
 ```
 
 --
@@ -144,4 +145,14 @@ This field supports xref:configuration:interpolation.adoc#bloblang-queries[inter
 *Type*: `string`
 
 
+=== `response_format`
+
+The format of the output, in one of these options: json, text, srt, verbose_json, or vtt.
+This field supports xref:configuration:interpolation.adoc#bloblang-queries[interpolation functions].
+
+
+*Type*: `string`
+
+*Default*: `"json"`
+
 

internal/impl/openai/transcription_processor.go

@@ -12,6 +12,7 @@ import (
 	"bytes"
 	"context"
 	"fmt"
+	"strings"
 
 	"github.com/redpanda-data/benthos/v4/public/bloblang"
 	"github.com/redpanda-data/benthos/v4/public/service"
@@ -22,6 +23,7 @@ const (
 	otspFieldFile   = "file"
 	otspFieldLang   = "language"
 	otspFieldPrompt = "prompt"
+	otspFieldFormat = "response_format"
 )
 
 func init() {
@@ -61,6 +63,11 @@ To learn more about audio transcription, see the: https://platform.openai.com/do
 				Description("Optional text to guide the model's style or continue a previous audio segment. The prompt should match the audio language.").
 				Optional().
 				Advanced(),
+			service.NewInterpolatedStringField(otspFieldFormat).
+				Description("The format of the output, in one of these options: json, text, srt, verbose_json, or vtt.").
+				Optional().
+				Default("json").
+				Advanced(),
 		)
 }
 
@@ -87,7 +94,14 @@ func makeTranscriptionProcessor(conf *service.ParsedConfig, mgr *service.Resourc
 			return nil, err
 		}
 	}
-	return &transcriptionProcessor{b, f, l, p}, nil
+	var t *service.InterpolatedString
+	if conf.Contains(otspFieldFormat) {
+		t, err = conf.FieldInterpolatedString(otspFieldFormat)
+		if err != nil {
+			return nil, err
+		}
+	}
+	return &transcriptionProcessor{b, f, l, p, t}, nil
 }
 
 type transcriptionProcessor struct {
@@ -96,6 +110,7 @@ type transcriptionProcessor struct {
 	file   *bloblang.Executor
 	lang   *service.InterpolatedString
 	prompt *service.InterpolatedString
+	format *service.InterpolatedString
 }
 
 func (p *transcriptionProcessor) Process(ctx context.Context, msg *service.Message) (service.MessageBatch, error) {
@@ -124,6 +139,30 @@ func (p *transcriptionProcessor) Process(ctx context.Context, msg *service.Messa
 		}
 		body.Prompt = pr
 	}
+
+	if p.format != nil {
+		t, err := p.format.TryString(msg)
+		if err != nil {
+			return nil, fmt.Errorf("%s interpolation error: %w", otspFieldFormat, err)
+		}
+
+		var format oai.AudioResponseFormat
+		switch strings.ToLower(t) {
+		case "verbose_json":
+			format = oai.AudioResponseFormatVerboseJSON
+		case "text":
+			format = oai.AudioResponseFormatText
+		case "vtt":
+			format = oai.AudioResponseFormatVTT
+		case "srt":
+			format = oai.AudioResponseFormatSRT
+		default:
+			format = oai.AudioResponseFormatJSON
+		}
+
+		body.Format = format
+	}
+
 	resp, err := p.client.CreateTranscription(ctx, body)
 	if err != nil {
 		return nil, err