@ai-matrx/media 0.12.0 → 0.14.0

package/dist/voices.d.cts

@@ -75,4 +75,119 @@ declare function voiceDisplayName(set: VoiceSetId, value: unknown): string;
 /** True when `id` is one of the live voice conversation (xAI Realtime) voices. */
 declare function isLiveConversationVoice(id: string): boolean;
 
-export { ASSISTANT_VOICE_ID, CARTESIA_API_VERSION, LEGACY_DEFAULT_VOICE_ID, LIVE_CONVERSATION_SAMPLE_MODEL, LIVE_CONVERSATION_VOICES, READING_VOICE_ID, TTS_DEFAULT_SPEED, TTS_DEFAULT_VOLUME, TTS_MODEL_ID, TTS_PLAYBACK_BUFFER_SEC, TTS_SPEED_MAX, TTS_SPEED_MIN, TTS_STREAMING_BUFFER_SEC, type VoiceOption, type VoicePurpose, type VoiceSetId, allVoices, availableVoices, isLiveConversationVoice, resolveSpeed, resolveVoiceId, voiceDisplayName, voiceOptions, voiceSetDefaultLabel, voiceSetOf };
+/** A term→spoken-form substitution applied before any other speech parsing. */
+interface SpeechPronunciation {
+    from: string;
+    to: string;
+}
+interface ParseMarkdownToTextOptions {
+    /**
+     * Custom Dictionary pronunciation substitutions. Applied FIRST — before the
+     * built-in abbreviation/number normalization — so a user term like "AI" can
+     * supply its own spoken form instead of the generic "A I" readout. Pairs
+     * should be sorted longest-first by the caller so multi-word terms replace
+     * before substrings.
+     */
+    pronunciations?: SpeechPronunciation[];
+}
+/**
+ * Long-form meanings retained as reference data for search, display, or future
+ * accessibility features. TTS must not use these expansions: repeatedly
+ * reading "Application Programming Interface" for API is not natural speech.
+ */
+declare const COMMON_ABBREVIATION_EXPANSIONS: {
+    readonly AI: "Artificial Intelligence";
+    readonly API: "Application Programming Interface";
+    readonly HTTP: "Hypertext Transfer Protocol";
+    readonly HTTPS: "Hypertext Transfer Protocol Secure";
+    readonly URL: "Uniform Resource Locator";
+    readonly URI: "Uniform Resource Identifier";
+    readonly JSON: "JavaScript Object Notation";
+    readonly XML: "eXtensible Markup Language";
+    readonly CSS: "Cascading Style Sheets";
+    readonly HTML: "Hypertext Markup Language";
+    readonly JS: "JavaScript";
+    readonly TS: "TypeScript";
+    readonly SQL: "Structured Query Language";
+    readonly DB: "Database";
+    readonly UI: "User Interface";
+    readonly UX: "User Experience";
+    readonly SEO: "Search Engine Optimization";
+    readonly SDK: "Software Development Kit";
+    readonly CLI: "Command Line Interface";
+    readonly IDE: "Integrated Development Environment";
+    readonly JWT: "JSON Web Token";
+    readonly OAuth: "Open Authorization";
+    readonly REST: "Representational State Transfer";
+    readonly CRUD: "Create Read Update Delete";
+    readonly MVC: "Model View Controller";
+    readonly SPA: "Single Page Application";
+    readonly SSR: "Server Side Rendering";
+    readonly CSR: "Client Side Rendering";
+    readonly PWA: "Progressive Web App";
+    readonly DOM: "Document Object Model";
+    readonly BOM: "Browser Object Model";
+    readonly CDN: "Content Delivery Network";
+    readonly CMS: "Content Management System";
+    readonly ERP: "Enterprise Resource Planning";
+    readonly CRM: "Customer Relationship Management";
+    readonly SaaS: "Software as a Service";
+    readonly PaaS: "Platform as a Service";
+    readonly IaaS: "Infrastructure as a Service";
+    readonly VPN: "Virtual Private Network";
+    readonly LAN: "Local Area Network";
+    readonly WAN: "Wide Area Network";
+    readonly TCP: "Transmission Control Protocol";
+    readonly UDP: "User Datagram Protocol";
+    readonly IP: "Internet Protocol";
+    readonly DNS: "Domain Name System";
+    readonly FTP: "File Transfer Protocol";
+    readonly SMTP: "Simple Mail Transfer Protocol";
+    readonly POP3: "Post Office Protocol version 3";
+    readonly IMAP: "Internet Message Access Protocol";
+};
+/**
+ * Normalize technical abbreviations for both active TTS lanes.
+ *
+ * Cartesia Sonic 3.5 recommends single-space character delimiters for forced
+ * readout ("A P I"), and that plain-text form also works with the catalog's
+ * Groq/Orpheus engine, which has no SSML/pronunciation-dictionary input. Word
+ * acronyms stay conventional so models can say "json", "oh-auth", etc.
+ */
+declare function normalizeSpeechAbbreviations(text: string): string;
+/**
+ * The spoken word a fill-in-the-blank run of underscores becomes. One constant
+ * so every speech surface reads the same thing — change it here (e.g. "what")
+ * and TTS + generated flashcard fronts follow.
+ */
+declare const SPEECH_BLANK_WORD = "blank";
+/**
+ * Fill-in-the-blank underscores → a spoken word. A run of two-or-more
+ * underscores (any length — "___", "_____"), including the spaced style
+ * ("_ _ _"), reads as "blank" instead of the literal "underscore" every TTS
+ * engine would otherwise say. Generic and reusable: the centralized speech
+ * cleaner AND the generated flashcard spoken-fronts both route through this, so
+ * a blank sounds identical everywhere.
+ *
+ * Guards (so it only ever touches real blanks):
+ *  - bounded by non-alphanumerics → never mangles `snake_case` or `__dunder__`;
+ *  - a line of ONLY underscores is a markdown thematic break (divider), left
+ *    untouched here for the horizontal-rule rule to drop.
+ */
+declare function normalizeSpeechBlanks(text: string): string;
+declare function parseMarkdownToText(markdown: string, options?: ParseMarkdownToTextOptions): string;
+
+/**
+ * The languages read-aloud offers, with each language's own name — the union of the website's
+ * voice-language setting and the extension's quick picker (which added Persian).
+ */
+interface ReadAloudLanguage {
+    code: string;
+    englishName: string;
+    nativeName: string;
+}
+declare const READ_ALOUD_LANGUAGES: readonly ReadAloudLanguage[];
+/** The language for `code`, or English when the code is not offered. */
+declare function readAloudLanguage(code: string | null | undefined): ReadAloudLanguage;
+
+export { ASSISTANT_VOICE_ID, CARTESIA_API_VERSION, COMMON_ABBREVIATION_EXPANSIONS, LEGACY_DEFAULT_VOICE_ID, LIVE_CONVERSATION_SAMPLE_MODEL, LIVE_CONVERSATION_VOICES, type ParseMarkdownToTextOptions, READING_VOICE_ID, READ_ALOUD_LANGUAGES, type ReadAloudLanguage, SPEECH_BLANK_WORD, type SpeechPronunciation, TTS_DEFAULT_SPEED, TTS_DEFAULT_VOLUME, TTS_MODEL_ID, TTS_PLAYBACK_BUFFER_SEC, TTS_SPEED_MAX, TTS_SPEED_MIN, TTS_STREAMING_BUFFER_SEC, type VoiceOption, type VoicePurpose, type VoiceSetId, allVoices, availableVoices, isLiveConversationVoice, normalizeSpeechAbbreviations, normalizeSpeechBlanks, parseMarkdownToText, readAloudLanguage, resolveSpeed, resolveVoiceId, voiceDisplayName, voiceOptions, voiceSetDefaultLabel, voiceSetOf };

package/dist/voices.d.ts

@@ -75,4 +75,119 @@ declare function voiceDisplayName(set: VoiceSetId, value: unknown): string;
 /** True when `id` is one of the live voice conversation (xAI Realtime) voices. */
 declare function isLiveConversationVoice(id: string): boolean;
 
-export { ASSISTANT_VOICE_ID, CARTESIA_API_VERSION, LEGACY_DEFAULT_VOICE_ID, LIVE_CONVERSATION_SAMPLE_MODEL, LIVE_CONVERSATION_VOICES, READING_VOICE_ID, TTS_DEFAULT_SPEED, TTS_DEFAULT_VOLUME, TTS_MODEL_ID, TTS_PLAYBACK_BUFFER_SEC, TTS_SPEED_MAX, TTS_SPEED_MIN, TTS_STREAMING_BUFFER_SEC, type VoiceOption, type VoicePurpose, type VoiceSetId, allVoices, availableVoices, isLiveConversationVoice, resolveSpeed, resolveVoiceId, voiceDisplayName, voiceOptions, voiceSetDefaultLabel, voiceSetOf };
+/** A term→spoken-form substitution applied before any other speech parsing. */
+interface SpeechPronunciation {
+    from: string;
+    to: string;
+}
+interface ParseMarkdownToTextOptions {
+    /**
+     * Custom Dictionary pronunciation substitutions. Applied FIRST — before the
+     * built-in abbreviation/number normalization — so a user term like "AI" can
+     * supply its own spoken form instead of the generic "A I" readout. Pairs
+     * should be sorted longest-first by the caller so multi-word terms replace
+     * before substrings.
+     */
+    pronunciations?: SpeechPronunciation[];
+}
+/**
+ * Long-form meanings retained as reference data for search, display, or future
+ * accessibility features. TTS must not use these expansions: repeatedly
+ * reading "Application Programming Interface" for API is not natural speech.
+ */
+declare const COMMON_ABBREVIATION_EXPANSIONS: {
+    readonly AI: "Artificial Intelligence";
+    readonly API: "Application Programming Interface";
+    readonly HTTP: "Hypertext Transfer Protocol";
+    readonly HTTPS: "Hypertext Transfer Protocol Secure";
+    readonly URL: "Uniform Resource Locator";
+    readonly URI: "Uniform Resource Identifier";
+    readonly JSON: "JavaScript Object Notation";
+    readonly XML: "eXtensible Markup Language";
+    readonly CSS: "Cascading Style Sheets";
+    readonly HTML: "Hypertext Markup Language";
+    readonly JS: "JavaScript";
+    readonly TS: "TypeScript";
+    readonly SQL: "Structured Query Language";
+    readonly DB: "Database";
+    readonly UI: "User Interface";
+    readonly UX: "User Experience";
+    readonly SEO: "Search Engine Optimization";
+    readonly SDK: "Software Development Kit";
+    readonly CLI: "Command Line Interface";
+    readonly IDE: "Integrated Development Environment";
+    readonly JWT: "JSON Web Token";
+    readonly OAuth: "Open Authorization";
+    readonly REST: "Representational State Transfer";
+    readonly CRUD: "Create Read Update Delete";
+    readonly MVC: "Model View Controller";
+    readonly SPA: "Single Page Application";
+    readonly SSR: "Server Side Rendering";
+    readonly CSR: "Client Side Rendering";
+    readonly PWA: "Progressive Web App";
+    readonly DOM: "Document Object Model";
+    readonly BOM: "Browser Object Model";
+    readonly CDN: "Content Delivery Network";
+    readonly CMS: "Content Management System";
+    readonly ERP: "Enterprise Resource Planning";
+    readonly CRM: "Customer Relationship Management";
+    readonly SaaS: "Software as a Service";
+    readonly PaaS: "Platform as a Service";
+    readonly IaaS: "Infrastructure as a Service";
+    readonly VPN: "Virtual Private Network";
+    readonly LAN: "Local Area Network";
+    readonly WAN: "Wide Area Network";
+    readonly TCP: "Transmission Control Protocol";
+    readonly UDP: "User Datagram Protocol";
+    readonly IP: "Internet Protocol";
+    readonly DNS: "Domain Name System";
+    readonly FTP: "File Transfer Protocol";
+    readonly SMTP: "Simple Mail Transfer Protocol";
+    readonly POP3: "Post Office Protocol version 3";
+    readonly IMAP: "Internet Message Access Protocol";
+};
+/**
+ * Normalize technical abbreviations for both active TTS lanes.
+ *
+ * Cartesia Sonic 3.5 recommends single-space character delimiters for forced
+ * readout ("A P I"), and that plain-text form also works with the catalog's
+ * Groq/Orpheus engine, which has no SSML/pronunciation-dictionary input. Word
+ * acronyms stay conventional so models can say "json", "oh-auth", etc.
+ */
+declare function normalizeSpeechAbbreviations(text: string): string;
+/**
+ * The spoken word a fill-in-the-blank run of underscores becomes. One constant
+ * so every speech surface reads the same thing — change it here (e.g. "what")
+ * and TTS + generated flashcard fronts follow.
+ */
+declare const SPEECH_BLANK_WORD = "blank";
+/**
+ * Fill-in-the-blank underscores → a spoken word. A run of two-or-more
+ * underscores (any length — "___", "_____"), including the spaced style
+ * ("_ _ _"), reads as "blank" instead of the literal "underscore" every TTS
+ * engine would otherwise say. Generic and reusable: the centralized speech
+ * cleaner AND the generated flashcard spoken-fronts both route through this, so
+ * a blank sounds identical everywhere.
+ *
+ * Guards (so it only ever touches real blanks):
+ *  - bounded by non-alphanumerics → never mangles `snake_case` or `__dunder__`;
+ *  - a line of ONLY underscores is a markdown thematic break (divider), left
+ *    untouched here for the horizontal-rule rule to drop.
+ */
+declare function normalizeSpeechBlanks(text: string): string;
+declare function parseMarkdownToText(markdown: string, options?: ParseMarkdownToTextOptions): string;
+
+/**
+ * The languages read-aloud offers, with each language's own name — the union of the website's
+ * voice-language setting and the extension's quick picker (which added Persian).
+ */
+interface ReadAloudLanguage {
+    code: string;
+    englishName: string;
+    nativeName: string;
+}
+declare const READ_ALOUD_LANGUAGES: readonly ReadAloudLanguage[];
+/** The language for `code`, or English when the code is not offered. */
+declare function readAloudLanguage(code: string | null | undefined): ReadAloudLanguage;
+
+export { ASSISTANT_VOICE_ID, CARTESIA_API_VERSION, COMMON_ABBREVIATION_EXPANSIONS, LEGACY_DEFAULT_VOICE_ID, LIVE_CONVERSATION_SAMPLE_MODEL, LIVE_CONVERSATION_VOICES, type ParseMarkdownToTextOptions, READING_VOICE_ID, READ_ALOUD_LANGUAGES, type ReadAloudLanguage, SPEECH_BLANK_WORD, type SpeechPronunciation, TTS_DEFAULT_SPEED, TTS_DEFAULT_VOLUME, TTS_MODEL_ID, TTS_PLAYBACK_BUFFER_SEC, TTS_SPEED_MAX, TTS_SPEED_MIN, TTS_STREAMING_BUFFER_SEC, type VoiceOption, type VoicePurpose, type VoiceSetId, allVoices, availableVoices, isLiveConversationVoice, normalizeSpeechAbbreviations, normalizeSpeechBlanks, parseMarkdownToText, readAloudLanguage, resolveSpeed, resolveVoiceId, voiceDisplayName, voiceOptions, voiceSetDefaultLabel, voiceSetOf };