Adding paragraphs feature to SDK

Author: michaeljolley

package-lock.json

@@ -0,0 +1,21 @@
+import { ParagraphGroup } from "./paragraphGroup";
+import { WordBase } from "./wordBase";
+
+export type Alternative = {
+  /**
+   * Text of speech identified by API
+   */
+  transcript: string;
+  /**
+   * Confidence in transcript generated
+   */
+  confidence: number;
+  /**
+   * Array of words included in the transcript
+   */
+  words: Array<WordBase>;
+  /**
+   * Array of paragraph objects.
+   */
+  paragraphs?: Array<ParagraphGroup>;
+};

src/types/alternatives.ts

@@ -1,5 +1,5 @@
+import { Alternative } from "./alternatives";
 import { Search } from "./search";
-import { WordBase } from "./wordBase";
 
 /**
  * Channel of speech identified by Deepgram
@@ -9,18 +9,8 @@ export type Channel = {
    * Searched terms & results
    */
   search?: Array<Search>;
-  alternatives: Array<{
-    /**
-     * Text of speech identified by API
-     */
-    transcript: string;
-    /**
-     * Confidence in transcript generated
-     */
-    confidence: number;
-    /**
-     * Array of words included in the transcript
-     */
-    words: Array<WordBase>;
-  }>;
+  /**
+   * Transcription alternatives
+   */
+  alternatives: Array<Alternative>;
 };

src/types/channel.ts

@@ -1,3 +1,4 @@
+export * from "./alternatives";
 export * from "./balance";
 export * from "./balanceList";
 export * from "./channel";
@@ -13,13 +14,16 @@ export * from "./member";
 export * from "./memberList";
 export * from "./message";
 export * from "./metadata";
+export * from "./paragraph";
+export * from "./paragraphGroup";
 export * from "./prerecordedTranscriptionOptions";
 export * from "./prerecordedTranscriptionResponse";
 export * from "./project";
 export * from "./projectPatchResponse";
 export * from "./projectResponse";
 export * from "./scopeList";
 export * from "./search";
+export * from "./sentence";
 export * from "./transcriptionSource";
 export * from "./usageCallback";
 export * from "./usageField";

src/types/index.ts

@@ -175,4 +175,11 @@ export type LiveTranscriptionOptions = {
    * @see https://developers.deepgram.com/api-reference/speech-recognition-api#operation/transcribeStreamingAudio/properties/sample_rate
    */
   sample_rate?: number;
+
+  /**
+   * Indicates whether Deepgram will split audio into paragraphs to improve transcript
+   * readability. When paragraphs is set to true, you must also set either punctuate,
+   * diarize, or multichannel to true.
+   */
+  paragraphs?: boolean;
 };

src/types/liveTranscriptionOptions.ts

@@ -0,0 +1,20 @@
+import { Sentence } from "./sentence";
+
+export type Paragraph = {
+  /**
+   * Sentences within the paragraph.
+   */
+  sentences: Array<Sentence>;
+  /**
+   * Start time (in seconds) from the start of the audio to where the paragraph starts.
+   */
+  start: number;
+  /**
+   * End time (in seconds) from the start of the audio to where the paragraph ends.
+   */
+  end: number;
+  /**
+   * Number of words in the paragraph
+   */
+  num_words: number;
+};

src/types/paragraph.ts

@@ -0,0 +1,12 @@
+import { Paragraph } from "./paragraph";
+
+export type ParagraphGroup = {
+  /**
+   * Full transcript
+   */
+  transcript: string;
+  /**
+   * Array of Paragraph objects.
+   */
+  paragraphs: Array<Paragraph>;
+};

src/types/paragraphGroup.ts

@@ -153,4 +153,11 @@ export type PrerecordedTranscriptionOptions = {
    * @see https://developers.deepgram.com/api-reference/speech-recognition-api#operation/transcribeAudio/properties/utt_split
    */
   utt_split?: number;
+
+  /**
+   * Indicates whether Deepgram will split audio into paragraphs to improve transcript
+   * readability. When paragraphs is set to true, you must also set either punctuate,
+   * diarize, or multichannel to true.
+   */
+  paragraphs?: boolean;
 };

src/types/prerecordedTranscriptionOptions.ts

@@ -0,0 +1,14 @@
+export type Sentence = {
+  /**
+   * Text transcript of the sentence.
+   */
+  text: string;
+  /**
+   * Start time (in seconds) from the start of the audio to where the sentence starts.
+   */
+  start: number;
+  /**
+   * End time (in seconds) from the start of the audio to where the sentence ends.
+   */
+  end: number;
+};

src/types/sentence.ts

@@ -1,15 +1,18 @@
 /*
-* The following const is replaced during the CD
-* process to the version from package.json
-*/
+ * The following const is replaced during the CD
+ * process to the version from package.json
+ */
 const sdkVersion = "DG_SDK_VERSION";
 
 export function userAgent(): string {
   let agent = "@deepgram/sdk/UNKNOWN node/UNKNOWN";
   try {
-    agent = `@deepgram/sdk/${sdkVersion} node/${process.version.replace("v", "")}`;
+    agent = `@deepgram/sdk/${sdkVersion} node/${process.version.replace(
+      "v",
+      ""
+    )}`;
   } catch (e) {
     console.warn("Could not load package details");
   }
   return agent;
-}
+}