@ainative/ai-kit-video 0.1.0

package/dist/processing/index.d.ts

@@ -0,0 +1,292 @@
+/**
+ * Timestamp granularity options for Whisper transcription
+ */
+type TimestampGranularity = 'word' | 'segment';
+/**
+ * Response format options for Whisper transcription
+ */
+type ResponseFormat = 'json' | 'text' | 'srt' | 'verbose_json' | 'vtt';
+/**
+ * Transcription segment with timing information
+ */
+interface TranscriptionSegment$1 {
+    id: number;
+    start: number;
+    end: number;
+    text: string;
+}
+/**
+ * Word-level timestamp information
+ */
+interface TranscriptionWord {
+    word: string;
+    start: number;
+    end: number;
+}
+/**
+ * Options for Whisper audio transcription
+ */
+interface TranscriptionOptions {
+    /**
+     * OpenAI API key for authentication
+     */
+    apiKey: string;
+    /**
+     * The language of the input audio in ISO-639-1 format (e.g., 'en', 'fr', 'es').
+     * Providing this can improve accuracy and latency.
+     */
+    language?: string;
+    /**
+     * Optional text to guide the model's style or continue a previous audio segment.
+     * The prompt should match the audio language and can be used for:
+     * - Maintaining consistent terminology (e.g., "Dr. Smith, AI, NLP")
+     * - Speaker identification hints (e.g., "Speaker 1, Speaker 2")
+     * - Ensuring proper punctuation and formatting
+     */
+    prompt?: string;
+    /**
+     * The format of the transcript output.
+     * - 'json': Returns basic JSON with text only
+     * - 'text': Returns plain text
+     * - 'srt': Returns SRT subtitle format
+     * - 'verbose_json': Returns detailed JSON with timestamps and metadata
+     * - 'vtt': Returns WebVTT subtitle format
+     */
+    response_format?: ResponseFormat;
+    /**
+     * 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 deterministic.
+     */
+    temperature?: number;
+    /**
+     * The timestamp granularities to populate for this transcription.
+     * Can include 'word' and/or 'segment' level timestamps.
+     * Note: There is no additional latency for segment timestamps, but word timestamps
+     * incur latency.
+     */
+    timestamp_granularities?: TimestampGranularity[];
+    /**
+     * The ID of the model to use. Only 'whisper-1' is currently available.
+     */
+    model?: string;
+}
+/**
+ * Result of audio transcription
+ */
+interface TranscriptionResult {
+    /**
+     * The transcribed text
+     */
+    text: string;
+    /**
+     * The language of the transcription (only available with verbose_json)
+     */
+    language?: string;
+    /**
+     * Duration of the audio in seconds (only available with verbose_json)
+     */
+    duration?: number;
+    /**
+     * Segment-level timestamps (only available when timestamp_granularities includes 'segment')
+     */
+    segments?: TranscriptionSegment$1[];
+    /**
+     * Word-level timestamps (only available when timestamp_granularities includes 'word')
+     */
+    words?: TranscriptionWord[];
+}
+/**
+ * Transcribe audio using OpenAI's Whisper API
+ *
+ * @param audioFile - The audio file to transcribe (File or Blob)
+ * @param options - Transcription options including API key and Whisper parameters
+ * @returns Promise resolving to the transcription result
+ *
+ * @throws {Error} When API key is missing or API call fails
+ */
+declare function transcribeAudio(audioFile: File | Blob, options: TranscriptionOptions): Promise<TranscriptionResult>;
+/**
+ * Utility function to format transcription segments into a readable format
+ *
+ * @param segments - Array of transcription segments with timestamps
+ * @returns Formatted string with timestamps
+ */
+declare function formatSegments(segments?: TranscriptionSegment$1[]): string;
+/**
+ * Utility function to extract speaker-labeled text from transcription
+ * Note: Whisper doesn't natively support speaker diarization, but you can
+ * guide it using prompts like "Speaker 1, Speaker 2" to encourage labeling.
+ *
+ * @param text - Transcribed text potentially containing speaker labels
+ * @returns Array of speaker segments
+ */
+declare function extractSpeakers(text: string): Array<{
+    speaker: string;
+    text: string;
+}>;
+/**
+ * Estimate the cost of transcribing audio based on duration
+ * Whisper pricing: $0.006 per minute
+ *
+ * @param durationSeconds - Duration of audio in seconds
+ * @returns Estimated cost in USD
+ */
+declare function estimateTranscriptionCost(durationSeconds: number): number;
+
+/**
+ * Transcription segment with timing
+ */
+interface TranscriptionSegment {
+    /** Transcribed text */
+    text: string;
+    /** Start timestamp in seconds */
+    start: number;
+    /** End timestamp in seconds */
+    end: number;
+    /** Confidence score (0-1) */
+    confidence?: number;
+    /** Speaker identifier */
+    speaker?: string;
+}
+/**
+ * Text formatting options
+ */
+interface TextFormattingOptions {
+    /** Enable automatic punctuation */
+    enablePunctuation?: boolean;
+    /** Enable capitalization */
+    enableCapitalization?: boolean;
+    /** Enable paragraph detection */
+    enableParagraphs?: boolean;
+    /** Maximum line length */
+    maxLineLength?: number;
+}
+
+/**
+ * Text Formatter for Transcription Processing
+ * @module @ainative/video/processing
+ */
+
+/**
+ * TextFormatter class for applying formatting to transcribed text
+ *
+ * Supports:
+ * - Automatic punctuation
+ * - Capitalization
+ * - Paragraph detection
+ * - Line length management
+ *
+ * @example
+ * ```typescript
+ * const formatter = new TextFormatter({
+ *   enablePunctuation: true,
+ *   enableCapitalization: true
+ * });
+ *
+ * const formatted = formatter.formatText('hello world');
+ * // Result: "Hello world."
+ * ```
+ */
+declare class TextFormatter {
+    private options;
+    /**
+     * Question word patterns for detecting questions
+     */
+    private static readonly QUESTION_WORDS;
+    /**
+     * Default formatting options
+     */
+    private static readonly DEFAULT_OPTIONS;
+    /**
+     * Creates a new TextFormatter instance
+     *
+     * @param options - Formatting options
+     */
+    constructor(options?: TextFormattingOptions);
+    /**
+     * Apply punctuation to text
+     *
+     * Adds appropriate punctuation marks based on sentence structure:
+     * - Periods for statements
+     * - Question marks for questions
+     * - Preserves existing punctuation
+     *
+     * @param text - Input text
+     * @returns Text with punctuation applied
+     */
+    applyPunctuation(text: string): string;
+    /**
+     * Apply capitalization to text
+     *
+     * Capitalizes:
+     * - First letter of text
+     * - Letters after sentence-ending punctuation
+     *
+     * @param text - Input text
+     * @returns Text with capitalization applied
+     */
+    applyCapitalization(text: string): string;
+    /**
+     * Format a single text string with all enabled options
+     *
+     * @param text - Input text
+     * @returns Formatted text
+     */
+    formatText(text: string): string;
+    /**
+     * Format a single line of text
+     *
+     * @param text - Input text (single line)
+     * @returns Formatted text
+     */
+    private formatSingleLine;
+    /**
+     * Format an array of transcription segments
+     *
+     * Applies formatting to the text of each segment while preserving
+     * timestamps, confidence scores, and speaker information.
+     *
+     * @param segments - Array of transcription segments
+     * @returns Array of formatted segments
+     */
+    formatSegments(segments: TranscriptionSegment[]): TranscriptionSegment[];
+    /**
+     * Get current formatting options
+     *
+     * @returns Current options
+     */
+    getOptions(): Readonly<Required<TextFormattingOptions>>;
+    /**
+     * Update formatting options
+     *
+     * @param options - New options to merge with existing
+     */
+    setOptions(options: Partial<TextFormattingOptions>): void;
+    /**
+     * Check if punctuation is enabled
+     *
+     * @returns True if punctuation is enabled
+     */
+    isPunctuationEnabled(): boolean;
+    /**
+     * Check if capitalization is enabled
+     *
+     * @returns True if capitalization is enabled
+     */
+    isCapitalizationEnabled(): boolean;
+    /**
+     * Check if paragraph detection is enabled
+     *
+     * @returns True if paragraph detection is enabled
+     */
+    isParagraphsEnabled(): boolean;
+    /**
+     * Get maximum line length
+     *
+     * @returns Maximum line length in characters
+     */
+    getMaxLineLength(): number;
+}
+
+export { type ResponseFormat, TextFormatter, type TextFormattingOptions, type TimestampGranularity, type TranscriptionOptions, type TranscriptionResult, type TranscriptionSegment$1 as TranscriptionSegment, type TranscriptionWord, estimateTranscriptionCost, extractSpeakers, formatSegments, transcribeAudio };

package/dist/processing/index.js

@@ -0,0 +1,308 @@
+'use strict';
+
+var OpenAI = require('openai');
+
+function _interopDefault (e) { return e && e.__esModule ? e : { default: e }; }
+
+var OpenAI__default = /*#__PURE__*/_interopDefault(OpenAI);
+
+// src/processing/transcription.ts
+async function transcribeAudio(audioFile, options) {
+  if (!options.apiKey || options.apiKey.trim() === "") {
+    throw new Error("OpenAI API key is required");
+  }
+  const openai = new OpenAI__default.default({
+    apiKey: options.apiKey
+  });
+  try {
+    const params = {
+      file: audioFile,
+      model: options.model || "whisper-1"
+    };
+    if (options.language) {
+      params.language = options.language;
+    }
+    if (options.prompt) {
+      params.prompt = options.prompt;
+    }
+    if (options.response_format) {
+      params.response_format = options.response_format;
+    }
+    if (options.temperature !== void 0) {
+      params.temperature = options.temperature;
+    }
+    if (options.timestamp_granularities && options.timestamp_granularities.length > 0) {
+      params.timestamp_granularities = options.timestamp_granularities;
+    }
+    const response = await openai.audio.transcriptions.create(params);
+    if (typeof response === "string") {
+      return { text: response };
+    }
+    const result = {
+      text: response.text
+    };
+    if ("language" in response) {
+      result.language = response.language;
+    }
+    if ("duration" in response) {
+      result.duration = response.duration;
+    }
+    if ("segments" in response) {
+      result.segments = response.segments;
+    }
+    if ("words" in response) {
+      result.words = response.words;
+    }
+    return result;
+  } catch (error) {
+    if (error instanceof Error) {
+      throw error;
+    }
+    throw new Error("Failed to transcribe audio: Unknown error");
+  }
+}
+function formatSegments(segments) {
+  if (!segments || segments.length === 0) {
+    return "";
+  }
+  return segments.map((segment) => {
+    const start = formatTimestamp(segment.start);
+    const end = formatTimestamp(segment.end);
+    return `[${start} - ${end}] ${segment.text.trim()}`;
+  }).join("\n");
+}
+function formatTimestamp(seconds) {
+  const mins = Math.floor(seconds / 60);
+  const secs = seconds % 60;
+  return `${mins.toString().padStart(2, "0")}:${secs.toFixed(1).padStart(4, "0")}`;
+}
+function extractSpeakers(text) {
+  const speakerPattern = /(Speaker \d+|[A-Z][a-z]+ [A-Z][a-z]+):\s*([^.!?]+[.!?])/g;
+  const matches = [...text.matchAll(speakerPattern)];
+  return matches.filter((match) => match[1] && match[2]).map((match) => ({
+    speaker: match[1].trim(),
+    text: match[2].trim()
+  }));
+}
+function estimateTranscriptionCost(durationSeconds) {
+  const minutes = durationSeconds / 60;
+  const costPerMinute = 6e-3;
+  return minutes * costPerMinute;
+}
+
+// src/processing/text-formatter.ts
+var _TextFormatter = class _TextFormatter {
+  /**
+   * Creates a new TextFormatter instance
+   *
+   * @param options - Formatting options
+   */
+  constructor(options = {}) {
+    this.options = {
+      ..._TextFormatter.DEFAULT_OPTIONS,
+      ...options
+    };
+  }
+  /**
+   * Apply punctuation to text
+   *
+   * Adds appropriate punctuation marks based on sentence structure:
+   * - Periods for statements
+   * - Question marks for questions
+   * - Preserves existing punctuation
+   *
+   * @param text - Input text
+   * @returns Text with punctuation applied
+   */
+  applyPunctuation(text) {
+    if (!text) {
+      return "";
+    }
+    text = text.trim();
+    if (!text) {
+      return "";
+    }
+    if (/[.!?]$/.test(text)) {
+      return text;
+    }
+    const lowerText = text.toLowerCase();
+    const firstWord = lowerText.split(/\s+/)[0];
+    if (firstWord && _TextFormatter.QUESTION_WORDS.includes(firstWord)) {
+      return text + "?";
+    }
+    return text + ".";
+  }
+  /**
+   * Apply capitalization to text
+   *
+   * Capitalizes:
+   * - First letter of text
+   * - Letters after sentence-ending punctuation
+   *
+   * @param text - Input text
+   * @returns Text with capitalization applied
+   */
+  applyCapitalization(text) {
+    if (!text) {
+      return "";
+    }
+    let result = text.charAt(0).toUpperCase() + text.slice(1);
+    result = result.replace(/([.!?]\s+)([a-z])/g, (_match, punctuation, letter) => {
+      return punctuation + letter.toUpperCase();
+    });
+    return result;
+  }
+  /**
+   * Format a single text string with all enabled options
+   *
+   * @param text - Input text
+   * @returns Formatted text
+   */
+  formatText(text) {
+    if (!text) {
+      return "";
+    }
+    let result = text;
+    if (text.includes("\n")) {
+      const lines = text.split("\n");
+      result = lines.map((line) => this.formatSingleLine(line)).join("\n");
+      return result;
+    }
+    return this.formatSingleLine(result);
+  }
+  /**
+   * Format a single line of text
+   *
+   * @param text - Input text (single line)
+   * @returns Formatted text
+   */
+  formatSingleLine(text) {
+    if (!text || !text.trim()) {
+      return text;
+    }
+    let result = text;
+    if (this.options.enableCapitalization) {
+      result = this.applyCapitalization(result);
+    }
+    if (this.options.enablePunctuation) {
+      result = this.applyPunctuation(result);
+    }
+    return result;
+  }
+  /**
+   * Format an array of transcription segments
+   *
+   * Applies formatting to the text of each segment while preserving
+   * timestamps, confidence scores, and speaker information.
+   *
+   * @param segments - Array of transcription segments
+   * @returns Array of formatted segments
+   */
+  formatSegments(segments) {
+    if (!segments || segments.length === 0) {
+      return [];
+    }
+    return segments.map((segment) => {
+      const formattedText = this.formatText(segment.text);
+      return {
+        ...segment,
+        text: formattedText
+      };
+    });
+  }
+  /**
+   * Get current formatting options
+   *
+   * @returns Current options
+   */
+  getOptions() {
+    return { ...this.options };
+  }
+  /**
+   * Update formatting options
+   *
+   * @param options - New options to merge with existing
+   */
+  setOptions(options) {
+    this.options = {
+      ...this.options,
+      ...options
+    };
+  }
+  /**
+   * Check if punctuation is enabled
+   *
+   * @returns True if punctuation is enabled
+   */
+  isPunctuationEnabled() {
+    return this.options.enablePunctuation;
+  }
+  /**
+   * Check if capitalization is enabled
+   *
+   * @returns True if capitalization is enabled
+   */
+  isCapitalizationEnabled() {
+    return this.options.enableCapitalization;
+  }
+  /**
+   * Check if paragraph detection is enabled
+   *
+   * @returns True if paragraph detection is enabled
+   */
+  isParagraphsEnabled() {
+    return this.options.enableParagraphs;
+  }
+  /**
+   * Get maximum line length
+   *
+   * @returns Maximum line length in characters
+   */
+  getMaxLineLength() {
+    return this.options.maxLineLength;
+  }
+};
+/**
+ * Question word patterns for detecting questions
+ */
+_TextFormatter.QUESTION_WORDS = [
+  "how",
+  "what",
+  "when",
+  "where",
+  "who",
+  "whom",
+  "whose",
+  "why",
+  "which",
+  "can",
+  "could",
+  "would",
+  "should",
+  "will",
+  "do",
+  "does",
+  "did",
+  "is",
+  "are",
+  "was",
+  "were"
+];
+/**
+ * Default formatting options
+ */
+_TextFormatter.DEFAULT_OPTIONS = {
+  enablePunctuation: true,
+  enableCapitalization: true,
+  enableParagraphs: false,
+  maxLineLength: 80
+};
+var TextFormatter = _TextFormatter;
+
+exports.TextFormatter = TextFormatter;
+exports.estimateTranscriptionCost = estimateTranscriptionCost;
+exports.extractSpeakers = extractSpeakers;
+exports.formatSegments = formatSegments;
+exports.transcribeAudio = transcribeAudio;
+//# sourceMappingURL=index.js.map
+//# sourceMappingURL=index.js.map

package/dist/processing/index.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../src/processing/transcription.ts","../../src/processing/text-formatter.ts"],"names":["OpenAI"],"mappings":";;;;;;;;;AA8HA,eAAsB,eAAA,CACpB,WACA,OAAA,EAC8B;AAE9B,EAAA,IAAI,CAAC,OAAA,CAAQ,MAAA,IAAU,QAAQ,MAAA,CAAO,IAAA,OAAW,EAAA,EAAI;AACnD,IAAA,MAAM,IAAI,MAAM,4BAA4B,CAAA;AAAA,EAC9C;AAGA,EAAA,MAAM,MAAA,GAAS,IAAIA,uBAAA,CAAO;AAAA,IACxB,QAAQ,OAAA,CAAQ;AAAA,GACjB,CAAA;AAED,EAAA,IAAI;AAEF,IAAA,MAAM,MAAA,GAAoC;AAAA,MACxC,IAAA,EAAM,SAAA;AAAA,MACN,KAAA,EAAO,QAAQ,KAAA,IAAS;AAAA,KAC1B;AAGA,IAAA,IAAI,QAAQ,QAAA,EAAU;AACpB,MAAA,MAAA,CAAO,WAAW,OAAA,CAAQ,QAAA;AAAA,IAC5B;AAEA,IAAA,IAAI,QAAQ,MAAA,EAAQ;AAClB,MAAA,MAAA,CAAO,SAAS,OAAA,CAAQ,MAAA;AAAA,IAC1B;AAEA,IAAA,IAAI,QAAQ,eAAA,EAAiB;AAC3B,MAAA,MAAA,CAAO,kBAAkB,OAAA,CAAQ,eAAA;AAAA,IACnC;AAEA,IAAA,IAAI,OAAA,CAAQ,gBAAgB,KAAA,CAAA,EAAW;AACrC,MAAA,MAAA,CAAO,cAAc,OAAA,CAAQ,WAAA;AAAA,IAC/B;AAEA,IAAA,IAAI,OAAA,CAAQ,uBAAA,IAA2B,OAAA,CAAQ,uBAAA,CAAwB,SAAS,CAAA,EAAG;AACjF,MAAA,MAAA,CAAO,0BAA0B,OAAA,CAAQ,uBAAA;AAAA,IAC3C;AAGA,IAAA,MAAM,WAAW,MAAM,MAAA,CAAO,KAAA,CAAM,cAAA,CAAe,OAAO,MAAM,CAAA;AAGhE,IAAA,IAAI,OAAO,aAAa,QAAA,EAAU;AAEhC,MAAA,OAAO,EAAE,MAAM,QAAA,EAAS;AAAA,IAC1B;AAGA,IAAA,MAAM,MAAA,GAA8B;AAAA,MAClC,MAAO,QAAA,CAAiB;AAAA,KAC1B;AAGA,IAAA,IAAI,cAAc,QAAA,EAAU;AAC1B,MAAA,MAAA,CAAO,WAAY,QAAA,CAAiB,QAAA;AAAA,IACtC;AAEA,IAAA,IAAI,cAAc,QAAA,EAAU;AAC1B,MAAA,MAAA,CAAO,WAAY,QAAA,CAAiB,QAAA;AAAA,IACtC;AAEA,IAAA,IAAI,cAAc,QAAA,EAAU;AAC1B,MAAA,MAAA,CAAO,WAAY,QAAA,CAAiB,QAAA;AAAA,IACtC;AAEA,IAAA,IAAI,WAAW,QAAA,EAAU;AACvB,MAAA,MAAA,CAAO,QAAS,QAAA,CAAiB,KAAA;AAAA,IACnC;AAEA,IAAA,OAAO,MAAA;AAAA,EACT,SAAS,KAAA,EAAO;AAEd,IAAA,IAAI,iBAAiB,KAAA,EAAO;AAC1B,MAAA,MAAM,KAAA;AAAA,IACR;AACA,IAAA,MAAM,IAAI,MAAM,2CAA2C,CAAA;AAAA,EAC7D;AACF;AAQO,SAAS,eAAe,QAAA,EAA2C;AACxE,EAAA,IAAI,CAAC,QAAA,IAAY,QAAA,CAAS,MAAA,KAAW,CAAA,EAAG;AACtC,IAAA,OAAO,EAAA;AAAA,EACT;AAEA,EAAA,OAAO,QAAA,CACJ,GAAA,CAAI,CAAC,OAAA,KAAY;AAChB,IAAA,MAAM,KAAA,GAAQ,eAAA,CAAgB,OAAA,CAAQ,KAAK,CAAA;AAC3C,IAAA,MAAM,GAAA,GAAM,eAAA,CAAgB,OAAA,CAAQ,GAAG,CAAA;AACvC,IAAA,OAAO,CAAA,CAAA,EAAI,KAAK,CAAA,GAAA,EAAM,GAAG,KAAK,OAAA,CAAQ,IAAA,CAAK,MAAM,CAAA,CAAA;AAAA,EACnD,CAAC,CAAA,CACA,IAAA,CAAK,IAAI,CAAA;AACd;AAQA,SAAS,gBAAgB,OAAA,EAAyB;AAChD,EAAA,MAAM,IAAA,GAAO,IAAA,CAAK,KAAA,CAAM,OAAA,GAAU,EAAE,CAAA;AACpC,EAAA,MAAM,OAAO,OAAA,GAAU,EAAA;AACvB,EAAA,OAAO,GAAG,IAAA,CAAK,QAAA,EAAS,CAAE,QAAA,CAAS,GAAG,GAAG,CAAC,CAAA,CAAA,EAAI,IAAA,CAAK,QAAQ,CAAC,CAAA,CAAE,QAAA,CAAS,CAAA,EAAG,GAAG,CAAC,CAAA,CAAA;AAChF;AAUO,SAAS,gBAAgB,IAAA,EAAwD;AACtF,EAAA,MAAM,cAAA,GAAiB,0DAAA;AACvB,EAAA,MAAM,UAAU,CAAC,GAAG,IAAA,CAAK,QAAA,CAAS,cAAc,CAAC,CAAA;AAEjD,EAAA,OAAO,OAAA,CACJ,MAAA,CAAO,CAAC,KAAA,KAAU,KAAA,CAAM,CAAC,CAAA,IAAK,KAAA,CAAM,CAAC,CAAC,CAAA,CACtC,GAAA,CAAI,CAAC,KAAA,MAAW;AAAA,IACf,OAAA,EAAS,KAAA,CAAM,CAAC,CAAA,CAAG,IAAA,EAAK;AAAA,IACxB,IAAA,EAAM,KAAA,CAAM,CAAC,CAAA,CAAG,IAAA;AAAK,GACvB,CAAE,CAAA;AACN;AASO,SAAS,0BAA0B,eAAA,EAAiC;AACzE,EAAA,MAAM,UAAU,eAAA,GAAkB,EAAA;AAClC,EAAA,MAAM,aAAA,GAAgB,IAAA;AACtB,EAAA,OAAO,OAAA,GAAU,aAAA;AACnB;;;ACrPO,IAAM,cAAA,GAAN,MAAM,cAAA,CAAc;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EA2BzB,WAAA,CAAY,OAAA,GAAiC,EAAC,EAAG;AAC/C,IAAA,IAAA,CAAK,OAAA,GAAU;AAAA,MACb,GAAG,cAAA,CAAc,eAAA;AAAA,MACjB,GAAG;AAAA,KACL;AAAA,EACF;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAaA,iBAAiB,IAAA,EAAsB;AAErC,IAAA,IAAI,CAAC,IAAA,EAAM;AACT,MAAA,OAAO,EAAA;AAAA,IACT;AAGA,IAAA,IAAA,GAAO,KAAK,IAAA,EAAK;AAGjB,IAAA,IAAI,CAAC,IAAA,EAAM;AACT,MAAA,OAAO,EAAA;AAAA,IACT;AAGA,IAAA,IAAI,QAAA,CAAS,IAAA,CAAK,IAAI,CAAA,EAAG;AACvB,MAAA,OAAO,IAAA;AAAA,IACT;AAGA,IAAA,MAAM,SAAA,GAAY,KAAK,WAAA,EAAY;AACnC,IAAA,MAAM,SAAA,GAAY,SAAA,CAAU,KAAA,CAAM,KAAK,EAAE,CAAC,CAAA;AAE1C,IAAA,IAAI,SAAA,IAAa,cAAA,CAAc,cAAA,CAAe,QAAA,CAAS,SAAS,CAAA,EAAG;AACjE,MAAA,OAAO,IAAA,GAAO,GAAA;AAAA,IAChB;AAGA,IAAA,OAAO,IAAA,GAAO,GAAA;AAAA,EAChB;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAYA,oBAAoB,IAAA,EAAsB;AAExC,IAAA,IAAI,CAAC,IAAA,EAAM;AACT,MAAA,OAAO,EAAA;AAAA,IACT;AAGA,IAAA,IAAI,MAAA,GAAS,KAAK,MAAA,CAAO,CAAC,EAAE,WAAA,EAAY,GAAI,IAAA,CAAK,KAAA,CAAM,CAAC,CAAA;AAGxD,IAAA,MAAA,GAAS,OAAO,OAAA,CAAQ,oBAAA,EAAsB,CAAC,MAAA,EAAQ,aAAa,MAAA,KAAW;AAC7E,MAAA,OAAO,WAAA,GAAc,OAAO,WAAA,EAAY;AAAA,IAC1C,CAAC,CAAA;AAED,IAAA,OAAO,MAAA;AAAA,EACT;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQA,WAAW,IAAA,EAAsB;AAC/B,IAAA,IAAI,CAAC,IAAA,EAAM;AACT,MAAA,OAAO,EAAA;AAAA,IACT;AAEA,IAAA,IAAI,MAAA,GAAS,IAAA;AAGb,IAAA,IAAI,IAAA,CAAK,QAAA,CAAS,IAAI,CAAA,EAAG;AACvB,MAAA,MAAM,KAAA,GAAQ,IAAA,CAAK,KAAA,CAAM,IAAI,CAAA;AAC7B,MAAA,MAAA,GAAS,KAAA,CAAM,IAAI,CAAA,IAAA,KAAQ,IAAA,CAAK,iBAAiB,IAAI,CAAC,CAAA,CAAE,IAAA,CAAK,IAAI,CAAA;AACjE,MAAA,OAAO,MAAA;AAAA,IACT;AAEA,IAAA,OAAO,IAAA,CAAK,iBAAiB,MAAM,CAAA;AAAA,EACrC;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQQ,iBAAiB,IAAA,EAAsB;AAC7C,IAAA,IAAI,CAAC,IAAA,IAAQ,CAAC,IAAA,CAAK,MAAK,EAAG;AACzB,MAAA,OAAO,IAAA;AAAA,IACT;AAEA,IAAA,IAAI,MAAA,GAAS,IAAA;AAGb,IAAA,IAAI,IAAA,CAAK,QAAQ,oBAAA,EAAsB;AACrC,MAAA,MAAA,GAAS,IAAA,CAAK,oBAAoB,MAAM,CAAA;AAAA,IAC1C;AAGA,IAAA,IAAI,IAAA,CAAK,QAAQ,iBAAA,EAAmB;AAClC,MAAA,MAAA,GAAS,IAAA,CAAK,iBAAiB,MAAM,CAAA;AAAA,IACvC;AAEA,IAAA,OAAO,MAAA;AAAA,EACT;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAWA,eAAe,QAAA,EAA0D;AAEvE,IAAA,IAAI,CAAC,QAAA,IAAY,QAAA,CAAS,MAAA,KAAW,CAAA,EAAG;AACtC,MAAA,OAAO,EAAC;AAAA,IACV;AAEA,IAAA,OAAO,QAAA,CAAS,IAAI,CAAA,OAAA,KAAW;AAE7B,MAAA,MAAM,aAAA,GAAgB,IAAA,CAAK,UAAA,CAAW,OAAA,CAAQ,IAAI,CAAA;AAGlD,MAAA,OAAO;AAAA,QACL,GAAG,OAAA;AAAA,QACH,IAAA,EAAM;AAAA,OACR;AAAA,IACF,CAAC,CAAA;AAAA,EACH;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOA,UAAA,GAAwD;AACtD,IAAA,OAAO,EAAE,GAAG,IAAA,CAAK,OAAA,EAAQ;AAAA,EAC3B;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOA,WAAW,OAAA,EAA+C;AACxD,IAAA,IAAA,CAAK,OAAA,GAAU;AAAA,MACb,GAAG,IAAA,CAAK,OAAA;AAAA,MACR,GAAG;AAAA,KACL;AAAA,EACF;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOA,oBAAA,GAAgC;AAC9B,IAAA,OAAO,KAAK,OAAA,CAAQ,iBAAA;AAAA,EACtB;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOA,uBAAA,GAAmC;AACjC,IAAA,OAAO,KAAK,OAAA,CAAQ,oBAAA;AAAA,EACtB;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOA,mBAAA,GAA+B;AAC7B,IAAA,OAAO,KAAK,OAAA,CAAQ,gBAAA;AAAA,EACtB;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOA,gBAAA,GAA2B;AACzB,IAAA,OAAO,KAAK,OAAA,CAAQ,aAAA;AAAA,EACtB;AACF,CAAA;AAAA;AAAA;AAAA;AA3Oa,cAAA,CAMa,cAAA,GAAiB;AAAA,EACvC,KAAA;AAAA,EAAO,MAAA;AAAA,EAAQ,MAAA;AAAA,EAAQ,OAAA;AAAA,EAAS,KAAA;AAAA,EAAO,MAAA;AAAA,EAAQ,OAAA;AAAA,EAC/C,KAAA;AAAA,EAAO,OAAA;AAAA,EAAS,KAAA;AAAA,EAAO,OAAA;AAAA,EAAS,OAAA;AAAA,EAAS,QAAA;AAAA,EACzC,MAAA;AAAA,EAAQ,IAAA;AAAA,EAAM,MAAA;AAAA,EAAQ,KAAA;AAAA,EAAO,IAAA;AAAA,EAAM,KAAA;AAAA,EAAO,KAAA;AAAA,EAAO;AACnD,CAAA;AAAA;AAAA;AAAA;AAVW,cAAA,CAea,eAAA,GAAmD;AAAA,EACzE,iBAAA,EAAmB,IAAA;AAAA,EACnB,oBAAA,EAAsB,IAAA;AAAA,EACtB,gBAAA,EAAkB,KAAA;AAAA,EAClB,aAAA,EAAe;AACjB,CAAA;AApBK,IAAM,aAAA,GAAN","file":"index.js","sourcesContent":["import OpenAI from 'openai'\nimport type { TranscriptionCreateParams } from 'openai/resources/audio/transcriptions'\n\n/**\n * Timestamp granularity options for Whisper transcription\n */\nexport type TimestampGranularity = 'word' | 'segment'\n\n/**\n * Response format options for Whisper transcription\n */\nexport type ResponseFormat = 'json' | 'text' | 'srt' | 'verbose_json' | 'vtt'\n\n/**\n * Transcription segment with timing information\n */\nexport interface TranscriptionSegment {\n  id: number\n  start: number\n  end: number\n  text: string\n}\n\n/**\n * Word-level timestamp information\n */\nexport interface TranscriptionWord {\n  word: string\n  start: number\n  end: number\n}\n\n/**\n * Options for Whisper audio transcription\n */\nexport interface TranscriptionOptions {\n  /**\n   * OpenAI API key for authentication\n   */\n  apiKey: string\n\n  /**\n   * The language of the input audio in ISO-639-1 format (e.g., 'en', 'fr', 'es').\n   * Providing this can improve accuracy and latency.\n   */\n  language?: string\n\n  /**\n   * Optional text to guide the model's style or continue a previous audio segment.\n   * The prompt should match the audio language and can be used for:\n   * - Maintaining consistent terminology (e.g., \"Dr. Smith, AI, NLP\")\n   * - Speaker identification hints (e.g., \"Speaker 1, Speaker 2\")\n   * - Ensuring proper punctuation and formatting\n   */\n  prompt?: string\n\n  /**\n   * The format of the transcript output.\n   * - 'json': Returns basic JSON with text only\n   * - 'text': Returns plain text\n   * - 'srt': Returns SRT subtitle format\n   * - 'verbose_json': Returns detailed JSON with timestamps and metadata\n   * - 'vtt': Returns WebVTT subtitle format\n   */\n  response_format?: ResponseFormat\n\n  /**\n   * The sampling temperature, between 0 and 1.\n   * Higher values like 0.8 will make the output more random,\n   * while lower values like 0.2 will make it more focused and deterministic.\n   */\n  temperature?: number\n\n  /**\n   * The timestamp granularities to populate for this transcription.\n   * Can include 'word' and/or 'segment' level timestamps.\n   * Note: There is no additional latency for segment timestamps, but word timestamps\n   * incur latency.\n   */\n  timestamp_granularities?: TimestampGranularity[]\n\n  /**\n   * The ID of the model to use. Only 'whisper-1' is currently available.\n   */\n  model?: string\n}\n\n/**\n * Result of audio transcription\n */\nexport interface TranscriptionResult {\n  /**\n   * The transcribed text\n   */\n  text: string\n\n  /**\n   * The language of the transcription (only available with verbose_json)\n   */\n  language?: string\n\n  /**\n   * Duration of the audio in seconds (only available with verbose_json)\n   */\n  duration?: number\n\n  /**\n   * Segment-level timestamps (only available when timestamp_granularities includes 'segment')\n   */\n  segments?: TranscriptionSegment[]\n\n  /**\n   * Word-level timestamps (only available when timestamp_granularities includes 'word')\n   */\n  words?: TranscriptionWord[]\n}\n\n/**\n * Transcribe audio using OpenAI's Whisper API\n *\n * @param audioFile - The audio file to transcribe (File or Blob)\n * @param options - Transcription options including API key and Whisper parameters\n * @returns Promise resolving to the transcription result\n *\n * @throws {Error} When API key is missing or API call fails\n */\nexport async function transcribeAudio(\n  audioFile: File | Blob,\n  options: TranscriptionOptions\n): Promise<TranscriptionResult> {\n  // Validate API key\n  if (!options.apiKey || options.apiKey.trim() === '') {\n    throw new Error('OpenAI API key is required')\n  }\n\n  // Initialize OpenAI client\n  const openai = new OpenAI({\n    apiKey: options.apiKey,\n  })\n\n  try {\n    // Build transcription parameters\n    const params: TranscriptionCreateParams = {\n      file: audioFile as File,\n      model: options.model || 'whisper-1',\n    }\n\n    // Add optional parameters\n    if (options.language) {\n      params.language = options.language\n    }\n\n    if (options.prompt) {\n      params.prompt = options.prompt\n    }\n\n    if (options.response_format) {\n      params.response_format = options.response_format\n    }\n\n    if (options.temperature !== undefined) {\n      params.temperature = options.temperature\n    }\n\n    if (options.timestamp_granularities && options.timestamp_granularities.length > 0) {\n      params.timestamp_granularities = options.timestamp_granularities\n    }\n\n    // Call Whisper API\n    const response = await openai.audio.transcriptions.create(params)\n\n    // Parse and return result based on response type\n    if (typeof response === 'string') {\n      // For text, srt, vtt formats\n      return { text: response }\n    }\n\n    // For json and verbose_json formats\n    const result: TranscriptionResult = {\n      text: (response as any).text,\n    }\n\n    // Add optional fields if present (from verbose_json)\n    if ('language' in response) {\n      result.language = (response as any).language\n    }\n\n    if ('duration' in response) {\n      result.duration = (response as any).duration\n    }\n\n    if ('segments' in response) {\n      result.segments = (response as any).segments\n    }\n\n    if ('words' in response) {\n      result.words = (response as any).words\n    }\n\n    return result\n  } catch (error) {\n    // Re-throw with original error message\n    if (error instanceof Error) {\n      throw error\n    }\n    throw new Error('Failed to transcribe audio: Unknown error')\n  }\n}\n\n/**\n * Utility function to format transcription segments into a readable format\n *\n * @param segments - Array of transcription segments with timestamps\n * @returns Formatted string with timestamps\n */\nexport function formatSegments(segments?: TranscriptionSegment[]): string {\n  if (!segments || segments.length === 0) {\n    return ''\n  }\n\n  return segments\n    .map((segment) => {\n      const start = formatTimestamp(segment.start)\n      const end = formatTimestamp(segment.end)\n      return `[${start} - ${end}] ${segment.text.trim()}`\n    })\n    .join('\\n')\n}\n\n/**\n * Format seconds into MM:SS.s format\n *\n * @param seconds - Time in seconds\n * @returns Formatted timestamp string\n */\nfunction formatTimestamp(seconds: number): string {\n  const mins = Math.floor(seconds / 60)\n  const secs = seconds % 60\n  return `${mins.toString().padStart(2, '0')}:${secs.toFixed(1).padStart(4, '0')}`\n}\n\n/**\n * Utility function to extract speaker-labeled text from transcription\n * Note: Whisper doesn't natively support speaker diarization, but you can\n * guide it using prompts like \"Speaker 1, Speaker 2\" to encourage labeling.\n *\n * @param text - Transcribed text potentially containing speaker labels\n * @returns Array of speaker segments\n */\nexport function extractSpeakers(text: string): Array<{ speaker: string; text: string }> {\n  const speakerPattern = /(Speaker \\d+|[A-Z][a-z]+ [A-Z][a-z]+):\\s*([^.!?]+[.!?])/g\n  const matches = [...text.matchAll(speakerPattern)]\n\n  return matches\n    .filter((match) => match[1] && match[2])\n    .map((match) => ({\n      speaker: match[1]!.trim(),\n      text: match[2]!.trim(),\n    }))\n}\n\n/**\n * Estimate the cost of transcribing audio based on duration\n * Whisper pricing: $0.006 per minute\n *\n * @param durationSeconds - Duration of audio in seconds\n * @returns Estimated cost in USD\n */\nexport function estimateTranscriptionCost(durationSeconds: number): number {\n  const minutes = durationSeconds / 60\n  const costPerMinute = 0.006\n  return minutes * costPerMinute\n}\n","/**\n * Text Formatter for Transcription Processing\n * @module @ainative/video/processing\n */\n\nimport type { TranscriptionSegment, TextFormattingOptions } from './types';\n\n/**\n * TextFormatter class for applying formatting to transcribed text\n *\n * Supports:\n * - Automatic punctuation\n * - Capitalization\n * - Paragraph detection\n * - Line length management\n *\n * @example\n * ```typescript\n * const formatter = new TextFormatter({\n *   enablePunctuation: true,\n *   enableCapitalization: true\n * });\n *\n * const formatted = formatter.formatText('hello world');\n * // Result: \"Hello world.\"\n * ```\n */\nexport class TextFormatter {\n  private options: Required<TextFormattingOptions>;\n\n  /**\n   * Question word patterns for detecting questions\n   */\n  private static readonly QUESTION_WORDS = [\n    'how', 'what', 'when', 'where', 'who', 'whom', 'whose',\n    'why', 'which', 'can', 'could', 'would', 'should',\n    'will', 'do', 'does', 'did', 'is', 'are', 'was', 'were'\n  ];\n\n  /**\n   * Default formatting options\n   */\n  private static readonly DEFAULT_OPTIONS: Required<TextFormattingOptions> = {\n    enablePunctuation: true,\n    enableCapitalization: true,\n    enableParagraphs: false,\n    maxLineLength: 80\n  };\n\n  /**\n   * Creates a new TextFormatter instance\n   *\n   * @param options - Formatting options\n   */\n  constructor(options: TextFormattingOptions = {}) {\n    this.options = {\n      ...TextFormatter.DEFAULT_OPTIONS,\n      ...options\n    };\n  }\n\n  /**\n   * Apply punctuation to text\n   *\n   * Adds appropriate punctuation marks based on sentence structure:\n   * - Periods for statements\n   * - Question marks for questions\n   * - Preserves existing punctuation\n   *\n   * @param text - Input text\n   * @returns Text with punctuation applied\n   */\n  applyPunctuation(text: string): string {\n    // Handle null/undefined input\n    if (!text) {\n      return '';\n    }\n\n    // Trim whitespace\n    text = text.trim();\n\n    // Return empty string if only whitespace\n    if (!text) {\n      return '';\n    }\n\n    // Already has ending punctuation\n    if (/[.!?]$/.test(text)) {\n      return text;\n    }\n\n    // Check if it's a question\n    const lowerText = text.toLowerCase();\n    const firstWord = lowerText.split(/\\s+/)[0];\n\n    if (firstWord && TextFormatter.QUESTION_WORDS.includes(firstWord)) {\n      return text + '?';\n    }\n\n    // Default to period\n    return text + '.';\n  }\n\n  /**\n   * Apply capitalization to text\n   *\n   * Capitalizes:\n   * - First letter of text\n   * - Letters after sentence-ending punctuation\n   *\n   * @param text - Input text\n   * @returns Text with capitalization applied\n   */\n  applyCapitalization(text: string): string {\n    // Handle empty input\n    if (!text) {\n      return '';\n    }\n\n    // Capitalize first letter\n    let result = text.charAt(0).toUpperCase() + text.slice(1);\n\n    // Capitalize after sentence-ending punctuation\n    result = result.replace(/([.!?]\\s+)([a-z])/g, (_match, punctuation, letter) => {\n      return punctuation + letter.toUpperCase();\n    });\n\n    return result;\n  }\n\n  /**\n   * Format a single text string with all enabled options\n   *\n   * @param text - Input text\n   * @returns Formatted text\n   */\n  formatText(text: string): string {\n    if (!text) {\n      return '';\n    }\n\n    let result = text;\n\n    // Handle multiline text\n    if (text.includes('\\n')) {\n      const lines = text.split('\\n');\n      result = lines.map(line => this.formatSingleLine(line)).join('\\n');\n      return result;\n    }\n\n    return this.formatSingleLine(result);\n  }\n\n  /**\n   * Format a single line of text\n   *\n   * @param text - Input text (single line)\n   * @returns Formatted text\n   */\n  private formatSingleLine(text: string): string {\n    if (!text || !text.trim()) {\n      return text;\n    }\n\n    let result = text;\n\n    // Apply capitalization first\n    if (this.options.enableCapitalization) {\n      result = this.applyCapitalization(result);\n    }\n\n    // Apply punctuation\n    if (this.options.enablePunctuation) {\n      result = this.applyPunctuation(result);\n    }\n\n    return result;\n  }\n\n  /**\n   * Format an array of transcription segments\n   *\n   * Applies formatting to the text of each segment while preserving\n   * timestamps, confidence scores, and speaker information.\n   *\n   * @param segments - Array of transcription segments\n   * @returns Array of formatted segments\n   */\n  formatSegments(segments: TranscriptionSegment[]): TranscriptionSegment[] {\n    // Handle empty array\n    if (!segments || segments.length === 0) {\n      return [];\n    }\n\n    return segments.map(segment => {\n      // Format the text\n      const formattedText = this.formatText(segment.text);\n\n      // Return new segment with formatted text, preserving all other properties\n      return {\n        ...segment,\n        text: formattedText\n      };\n    });\n  }\n\n  /**\n   * Get current formatting options\n   *\n   * @returns Current options\n   */\n  getOptions(): Readonly<Required<TextFormattingOptions>> {\n    return { ...this.options };\n  }\n\n  /**\n   * Update formatting options\n   *\n   * @param options - New options to merge with existing\n   */\n  setOptions(options: Partial<TextFormattingOptions>): void {\n    this.options = {\n      ...this.options,\n      ...options\n    };\n  }\n\n  /**\n   * Check if punctuation is enabled\n   *\n   * @returns True if punctuation is enabled\n   */\n  isPunctuationEnabled(): boolean {\n    return this.options.enablePunctuation;\n  }\n\n  /**\n   * Check if capitalization is enabled\n   *\n   * @returns True if capitalization is enabled\n   */\n  isCapitalizationEnabled(): boolean {\n    return this.options.enableCapitalization;\n  }\n\n  /**\n   * Check if paragraph detection is enabled\n   *\n   * @returns True if paragraph detection is enabled\n   */\n  isParagraphsEnabled(): boolean {\n    return this.options.enableParagraphs;\n  }\n\n  /**\n   * Get maximum line length\n   *\n   * @returns Maximum line length in characters\n   */\n  getMaxLineLength(): number {\n    return this.options.maxLineLength;\n  }\n}\n"]}