@huggingface/transformers 3.0.0-alpha.0

package/src/models/whisper/common_whisper.js

@@ -0,0 +1,151 @@
+
+
+const WHISPER_LANGUAGES = [
+    ["en", "english"],
+    ["zh", "chinese"],
+    ["de", "german"],
+    ["es", "spanish"],
+    ["ru", "russian"],
+    ["ko", "korean"],
+    ["fr", "french"],
+    ["ja", "japanese"],
+    ["pt", "portuguese"],
+    ["tr", "turkish"],
+    ["pl", "polish"],
+    ["ca", "catalan"],
+    ["nl", "dutch"],
+    ["ar", "arabic"],
+    ["sv", "swedish"],
+    ["it", "italian"],
+    ["id", "indonesian"],
+    ["hi", "hindi"],
+    ["fi", "finnish"],
+    ["vi", "vietnamese"],
+    ["he", "hebrew"],
+    ["uk", "ukrainian"],
+    ["el", "greek"],
+    ["ms", "malay"],
+    ["cs", "czech"],
+    ["ro", "romanian"],
+    ["da", "danish"],
+    ["hu", "hungarian"],
+    ["ta", "tamil"],
+    ["no", "norwegian"],
+    ["th", "thai"],
+    ["ur", "urdu"],
+    ["hr", "croatian"],
+    ["bg", "bulgarian"],
+    ["lt", "lithuanian"],
+    ["la", "latin"],
+    ["mi", "maori"],
+    ["ml", "malayalam"],
+    ["cy", "welsh"],
+    ["sk", "slovak"],
+    ["te", "telugu"],
+    ["fa", "persian"],
+    ["lv", "latvian"],
+    ["bn", "bengali"],
+    ["sr", "serbian"],
+    ["az", "azerbaijani"],
+    ["sl", "slovenian"],
+    ["kn", "kannada"],
+    ["et", "estonian"],
+    ["mk", "macedonian"],
+    ["br", "breton"],
+    ["eu", "basque"],
+    ["is", "icelandic"],
+    ["hy", "armenian"],
+    ["ne", "nepali"],
+    ["mn", "mongolian"],
+    ["bs", "bosnian"],
+    ["kk", "kazakh"],
+    ["sq", "albanian"],
+    ["sw", "swahili"],
+    ["gl", "galician"],
+    ["mr", "marathi"],
+    ["pa", "punjabi"],
+    ["si", "sinhala"],
+    ["km", "khmer"],
+    ["sn", "shona"],
+    ["yo", "yoruba"],
+    ["so", "somali"],
+    ["af", "afrikaans"],
+    ["oc", "occitan"],
+    ["ka", "georgian"],
+    ["be", "belarusian"],
+    ["tg", "tajik"],
+    ["sd", "sindhi"],
+    ["gu", "gujarati"],
+    ["am", "amharic"],
+    ["yi", "yiddish"],
+    ["lo", "lao"],
+    ["uz", "uzbek"],
+    ["fo", "faroese"],
+    ["ht", "haitian creole"],
+    ["ps", "pashto"],
+    ["tk", "turkmen"],
+    ["nn", "nynorsk"],
+    ["mt", "maltese"],
+    ["sa", "sanskrit"],
+    ["lb", "luxembourgish"],
+    ["my", "myanmar"],
+    ["bo", "tibetan"],
+    ["tl", "tagalog"],
+    ["mg", "malagasy"],
+    ["as", "assamese"],
+    ["tt", "tatar"],
+    ["haw", "hawaiian"],
+    ["ln", "lingala"],
+    ["ha", "hausa"],
+    ["ba", "bashkir"],
+    ["jw", "javanese"],
+    ["su", "sundanese"],
+]
+
+// @ts-ignore
+export const WHISPER_LANGUAGE_MAPPING = new Map(WHISPER_LANGUAGES);
+// @ts-ignore
+export const WHISPER_TO_LANGUAGE_CODE_MAPPING = new Map([
+    ...WHISPER_LANGUAGES.map(([k, v]) => [v, k]),
+    ...[
+        ["burmese", "my"],
+        ["valencian", "ca"],
+        ["flemish", "nl"],
+        ["haitian", "ht"],
+        ["letzeburgesch", "lb"],
+        ["pushto", "ps"],
+        ["panjabi", "pa"],
+        ["moldavian", "ro"],
+        ["moldovan", "ro"],
+        ["sinhalese", "si"],
+        ["castilian", "es"],
+    ]
+]);
+
+/**
+ * @param {string} language The language name or code
+ * @returns {string} The language code
+ */
+export function whisper_language_to_code(language) {
+    language = language.toLowerCase();
+
+    // Map to code from user-friendly name (e.g., "english" -> "en")
+    let language_code = WHISPER_TO_LANGUAGE_CODE_MAPPING.get(language);
+
+    if (language_code === undefined) {
+        // User provided something that is not a language name
+
+        if (WHISPER_LANGUAGE_MAPPING.has(language)) {
+            // User provided the language code directly (e.g., "en")
+            language_code = language;
+
+        } else {
+            // User provided something that is not a language code or name
+            const is_language_code = language.length === 2;
+            const langs = is_language_code ? WHISPER_LANGUAGE_MAPPING.keys() : WHISPER_LANGUAGE_MAPPING.values();
+
+            throw new Error(`Language "${language}" is not supported. Must be one of: ${JSON.stringify(langs)}`);
+        }
+    }
+    return language_code;
+}

package/src/models/whisper/generation_whisper.js

@@ -0,0 +1,89 @@
+import { GenerationConfig } from "../../generation/configuration_utils.js";
+
+export class WhisperGenerationConfig extends GenerationConfig {
+
+    /**
+     * Whether to return the timestamps with the text. This enables the `WhisperTimestampsLogitsProcessor`.
+     * @type {boolean}
+     */
+    return_timestamps = null;
+
+    /**
+     * Whether to return token-level timestamps
+     * with the text. This can be used with or without the `return_timestamps` option. To get word-level
+     * timestamps, use the tokenizer to group the tokens into words.
+     * @type {boolean}
+     */
+    return_token_timestamps = null;
+
+    /**
+     * The number of audio frames available in this chunk. This is only used generating word-level timestamps.
+     * @type {number}
+     */
+    num_frames = null;
+
+    /**
+     * Alignment heads to predict word-level timestamps. This is a list of [layer, head] pairs that
+     * select the cross-attention heads that are highly correlated to word-level timing.
+     * @type {[number, number][]}
+     */
+    alignment_heads = null;
+
+    /**
+     * Task to use for generation, either "translate" or "transcribe".
+     * @type {string}
+     */
+    task = null;
+
+    /**
+     * Language token to use for generation, can be either in the form of `<|en|>`, `en` or `english`.
+     * You can find all the possible language tokens in the `model.generation_config.lang_to_id` dictionary.
+     * @type {string}
+     */
+    language = null;
+
+    /**
+     * The id of the `"<|notimestamps|>"` token.
+     * @type {number}
+     */
+    no_timestamps_token_id = null;
+
+    /**
+     * Rank-1 list of token IDs created by passing text to [`~WhisperProcessor.get_prompt_ids`] that is
+     * provided as a prompt to each chunk. This can be used to provide or "prompt-engineer" a context for
+     * transcription, e.g. custom vocabularies or proper nouns to make it more likely to predict those words
+     * correctly. It cannot be used in conjunction with `decoder_start_token_id` as it overwrites this value.
+     * @type {number[]}
+     */
+    prompt_ids = null;
+
+    /**
+     * Whether the model is multilingual or not.
+     * @type {boolean}
+     */
+    is_multilingual = null;
+
+    /**
+     * (Optional) A mapping from language tokens to their corresponding IDs.
+     * Only required if the model is multilingual.
+     * @type {Record<string, number>|null}
+     */
+    lang_to_id = null;
+
+    /**
+     * (Optional) A mapping from task tokens to their corresponding IDs.
+     * @type {Record<string, number>|null}
+     */
+    task_to_id = null;
+
+    /**
+     * Used to set the maximum value of the initial timestamp. This is used to prevent the model from
+     * predicting timestamps that are too far in the future.
+     * @type {number}
+     */
+    max_initial_timestamp_index = 1;
+}
+
+/**
+ * @typedef {import('../../generation/parameters.js').GenerationFunctionParameters & {generation_config: WhisperGenerationConfig} & WhisperGenerationConfig} WhisperGenerationFunctionParameters
+ */