@remotion/whisper-web 4.0.364 → 4.0.366

package/package.json

@@ -3,7 +3,7 @@
 		"url": "https://github.com/remotion-dev/remotion/tree/main/packages/whisper-web"
 	},
 	"name": "@remotion/whisper-web",
-	"version": "4.0.364",
+	"version": "4.0.366",
 	"main": "dist/index.js",
 	"sideEffects": false,
 	"scripts": {
@@ -14,10 +14,10 @@
 	"author": "Hunain Ahmed <junaidhunain6@gmail.com>",
 	"license": "UNLICENSED",
 	"dependencies": {
-		"@remotion/captions": "4.0.364"
+		"@remotion/captions": "4.0.366"
 	},
 	"devDependencies": {
-		"@remotion/eslint-config-internal": "4.0.364",
+		"@remotion/eslint-config-internal": "4.0.366",
 		"eslint": "9.19.0"
 	},
 	"publishConfig": {

package/.turbo/turbo-formatting.log

@@ -1,4 +0,0 @@
-
-$ prettier --experimental-cli src --check
-[?25l- Checking formatting...\ Checking formatting...| Checking formatting.../ Checking formatting...- Checking formatting...\ Checking formatting...| Checking formatting.../ Checking formatting...- Checking formatting...\ Checking formatting...| Checking formatting.../ Checking formatting...- Checking formatting...\ Checking formatting...| Checking formatting.../ Checking formatting...- Checking formatting...\ Checking formatting...| Checking formatting.../ Checking formatting...- Checking formatting...\ Checking formatting...| Checking formatting.../ Checking formatting...- Checking formatting...\ Checking formatting...| Checking formatting.../ Checking formatting...- Checking formatting...\ Checking formatting...| Checking formatting.../ Checking formatting...- Checking formatting...\ Checking formatting...| Checking formatting.../ Checking formatting...- Checking formatting...\ Checking formatting...| Checking formatting.../ Checking formatting...- Checking formatting...\ Checking formatting...| Checking formatting.../ Checking formatting...- Checking formatting...\ Checking formatting...| Checking formatting.../ Checking formatting...- Checking formatting...\ Checking formatting...| Checking formatting.../ src/get-loaded-models.ts- src/download-whisper-model.ts\ src/download-whisper-model.ts| src/download-whisper-model.ts/ src/simulate-progress.ts- src/download-model.ts\ src/download-model.ts| src/db/open-db.ts/ src/get-model-url.ts- src/db/get-object-from-db.ts\ src/db/get-object-from-db.ts| src/index.module.ts/ src/get-available-models.ts- src/can-use-whisper-web.ts\ src/resample-to-16khz.ts| src/delete-model.ts/ src/delete-model.ts[?25hChecking formatting...
-[?25hAll matched files use Prettier code style!

package/.turbo/turbo-lint.log

@@ -1,25 +0,0 @@
-
-$ eslint src
-=============
-
-WARNING: You are currently running a version of TypeScript which is not officially supported by @typescript-eslint/typescript-estree.
-
-You may find that it works just fine, or you may not.
-
-SUPPORTED TYPESCRIPT VERSIONS: >=4.7.4 <5.7.0
-
-YOUR TYPESCRIPT VERSION: 5.8.2
-
-Please only submit bug reports when using the officially supported version.
-
-=============
-
-/Users/jonathanburger/remotion/packages/whisper-web/src/resample-to-16khz.ts
-  47:50  warning  Async arrow function has no 'await' expression  require-await
-
-/Users/jonathanburger/remotion/packages/whisper-web/src/transcribe.ts
-  18:60  warning  Unexpected any. Specify a different type  @typescript-eslint/no-explicit-any
-  51:55  warning  Unexpected any. Specify a different type  @typescript-eslint/no-explicit-any
-
-✖ 3 problems (0 errors, 3 warnings)
-

package/.turbo/turbo-make.log

@@ -1,3 +0,0 @@
-
-$ tsc -d && bun --env-file=../.env.bundle bundle.ts
-[16.38ms] Generated.

package/build-wasm.ts

@@ -1,90 +0,0 @@
-// TODO: there's no indication of error in case worker.js is fails to dynamically import
-
-import {$} from 'bun';
-import fs from 'fs';
-import os from 'os';
-import path from 'path';
-
-const randomDir = os.tmpdir();
-
-const wasmDir = path.join(randomDir, 'whisper-web');
-
-if (fs.existsSync(wasmDir)) {
-	fs.rmSync(wasmDir, {recursive: true});
-}
-
-fs.mkdirSync(wasmDir, {recursive: true});
-
-const cwd = path.join(wasmDir, 'build-em');
-
-await $`git clone https://github.com/ggerganov/whisper.cpp ${wasmDir}`;
-await $`git checkout v1.7.5`.cwd(wasmDir);
-
-fs.mkdirSync(cwd);
-
-const cmakeListsFile = path.join(
-	wasmDir,
-	'examples',
-	'whisper.wasm',
-	'CMakeLists.txt',
-);
-
-const file = fs.readFileSync(cmakeListsFile, 'utf8');
-
-// Disable Node.JS target, compile with assertions to get stack trace
-const lines = file.split('\n').map((line) => {
-	if (line.includes('-s FORCE_FILESYSTEM=1 \\')) {
-		// output ES6 module so we can import it dynamically without injeting the script tag
-		return `-s FORCE_FILESYSTEM=1 -s ENVIRONMENT='web,worker' -s EXPORT_ES6=1 -s MODULARIZE=1 -s EXPORT_NAME=\\"createModule\\" \\`;
-	}
-
-	if (line.includes('-s EXPORTED_RUNTIME_METHODS')) {
-		return `-s EXPORTED_RUNTIME_METHODS=\\"['print', 'printErr', 'ccall', 'cwrap', 'HEAPU8']\\" \\`;
-	}
-
-	return line;
-});
-
-fs.writeFileSync(
-	cmakeListsFile,
-	[
-		...lines,
-		// Generate a .d.ts file
-		`set(CMAKE_CXX_FLAGS "$\{CMAKE_CXX_FLAGS\} -lembind --emit-tsd $\{CMAKE_CURRENT_BINARY_DIR\}/interface.d.ts")`,
-	].join('\n'),
-);
-
-const emscriptenFilePath = path.join(
-	wasmDir,
-	'examples',
-	'whisper.wasm',
-	'emscripten.cpp',
-);
-//now get our version
-const modifiedVersion = fs.readFileSync('./emscripten.cpp', 'utf8');
-fs.writeFileSync(emscriptenFilePath, modifiedVersion);
-
-// brew install emscripten if necessary
-await $`emcmake cmake ..`.cwd(cwd);
-await $`make -j`.cwd(cwd);
-
-const mainJsFile = path.join(cwd, 'bin', 'whisper.wasm', 'main.js');
-
-let content = fs
-	.readFileSync(mainJsFile, 'utf8')
-	.replace('libmain.js', './main.js');
-
-// Write the modified content directly to the destination
-fs.writeFileSync(path.join(__dirname, 'main.js'), content, 'utf8');
-
-const dTsFile = path.join(
-	wasmDir,
-	'build-em',
-	'examples',
-	'whisper.wasm',
-	'interface.d.ts',
-);
-
-fs.copyFileSync(dTsFile, path.join(__dirname, 'main.d.ts'));
-
-fs.rmSync(wasmDir, {recursive: true});

package/bundle.ts

@@ -1,15 +0,0 @@
-import {buildPackage} from '../.monorepo/builder';
-
-await buildPackage({
-	formats: {
-		cjs: 'use-tsc',
-		esm: 'build',
-	},
-	external: ['./main.js', './worker.js'],
-	entrypoints: [
-		{
-			path: 'src/index.module.ts',
-			target: 'browser',
-		},
-	],
-});

package/emscripten.cpp

@@ -1,303 +0,0 @@
-#include "whisper.h"
-
-#include <emscripten.h>
-#include <emscripten/bind.h>
-#include <iostream>
-#include <vector>
-#include <thread>
-
-std::thread g_worker;
-
-static inline int mpow2(int n) {
-    int p = 1;
-    while (p <= n) p *= 2;
-    return p / 2;
-}
-
-static char * escape_double_quotes_and_backslashes(const char * str) {
-    if (str == NULL) {
-        return NULL;
-    }
-
-    size_t escaped_length = strlen(str) + 1;
-
-    for (size_t i = 0; str[i] != '\0'; i++) {
-        if (str[i] == '"' || str[i] == '\\') {
-            escaped_length++;
-        }
-    }
-
-    char * escaped = (char *)calloc(escaped_length, 1); // pre-zeroed
-    if (escaped == NULL) {
-        return NULL;
-    }
-
-    size_t pos = 0;
-    for (size_t i = 0; str[i] != '\0'; i++) {
-        if (str[i] == '"' || str[i] == '\\') {
-            escaped[pos++] = '\\';
-        }
-        escaped[pos++] = str[i];
-    }
-
-    // no need to set zero due to calloc() being used prior
-
-    return escaped;
-}
-
-//  500 -> 00:05.000
-// 6000 -> 01:00.000
-std::string to_timestamp(int64_t t, bool comma) {
-    int64_t msec = t * 10;
-    int64_t hr = msec / (1000 * 60 * 60);
-    msec = msec - hr * (1000 * 60 * 60);
-    int64_t min = msec / (1000 * 60);
-    msec = msec - min * (1000 * 60);
-    int64_t sec = msec / 1000;
-    msec = msec - sec * 1000;
-
-    char buf[32];
-    snprintf(buf, sizeof(buf), "%02d:%02d:%02d%s%03d", (int) hr, (int) min, (int) sec, comma ? "," : ".", (int) msec);
-
-    return std::string(buf);
-}
-
-static bool output_json(
-             struct whisper_context * ctx,
-                               bool   final,
-                               int    first_segment,
-                               int    n_segments) {
-    int indent = 0;
-    std::string output;
-
-
-    auto doindent = [&]() {
-        for (int i = 0; i < indent; i++) output += "\t";
-    };
-
-    auto start_arr = [&](const char *name) {
-        doindent();
-        output += "\"" + std::string(name) + "\": [";
-        indent++;
-    };
-
-    auto end_arr = [&](bool end) {
-        indent--;
-        doindent();
-        output += (end ? "]" : "],");
-    };
-
-    auto start_obj = [&](const char *name) {
-        doindent();
-        if (name) {
-            output += "\"" + std::string(name) + "\": {";
-        } else {
-            output += "{";
-        }
-        indent++;
-    };
-
-    auto end_obj = [&](bool end) {
-        indent--;
-        doindent();
-        output += (end ? "}" : "},");
-    };
-
-    auto start_value = [&](const char *name) {
-        doindent();
-        output += "\"" + std::string(name) + "\": ";
-    };
-
-    auto value_s = [&](const char *name, const char *val, bool end) {
-        start_value(name);
-        char * val_escaped = escape_double_quotes_and_backslashes(val);
-        output += "\"" + std::string(val_escaped) + (end ? "\"" : "\",");
-        free(val_escaped);
-    };
-
-    auto end_value = [&](bool end) {
-        output += (end ? "" : ",");
-    };
-
-    auto value_i = [&](const char *name, const int64_t val, bool end) {
-        start_value(name);
-        output += std::to_string(val);
-        end_value(end);
-    };
-
-    auto value_f = [&](const char *name, const float val, bool end) {
-        start_value(name);
-        output += std::to_string(val);
-        end_value(end);
-    };
-
-    auto value_b = [&](const char *name, const bool val, bool end) {
-        start_value(name);
-        output += (val ? "true" : "false");
-        end_value(end);
-    };
-
-    auto times_o = [&](int64_t t0, int64_t t1, bool end) {
-        start_obj("timestamps");
-        value_s("from", to_timestamp(t0, true).c_str(), false);
-        value_s("to", to_timestamp(t1, true).c_str(), true);
-        end_obj(false);
-        start_obj("offsets");
-        value_i("from", t0 * 10, false);
-        value_i("to", t1 * 10, true);
-        end_obj(end);
-    };
-
-    start_obj(nullptr);
-        value_s("systeminfo", whisper_print_system_info(), false);
-        start_obj("model");
-            value_s("type", whisper_model_type_readable(ctx), false);
-            value_b("multilingual", whisper_is_multilingual(ctx), false);
-            value_i("vocab", whisper_model_n_vocab(ctx), false);
-            start_obj("audio");
-                value_i("ctx", whisper_model_n_audio_ctx(ctx), false);
-                value_i("state", whisper_model_n_audio_state(ctx), false);
-                value_i("head", whisper_model_n_audio_head(ctx), false);
-                value_i("layer", whisper_model_n_audio_layer(ctx), true);
-            end_obj(false);
-            start_obj("text");
-                value_i("ctx", whisper_model_n_text_ctx(ctx), false);
-                value_i("state", whisper_model_n_text_state(ctx), false);
-                value_i("head", whisper_model_n_text_head(ctx), false);
-                value_i("layer", whisper_model_n_text_layer(ctx), true);
-            end_obj(false);
-            value_i("mels", whisper_model_n_mels(ctx), false);
-            value_i("ftype", whisper_model_ftype(ctx), true);
-        end_obj(false);
-        start_obj("result");
-            value_s("language", whisper_lang_str(whisper_full_lang_id(ctx)), true);
-        end_obj(false);
-        start_arr("transcription");
-
-        for (int i = first_segment; i < n_segments; ++i) {
-            const char * text = whisper_full_get_segment_text(ctx, i);
-
-            const int64_t t0 = whisper_full_get_segment_t0(ctx, i);
-            const int64_t t1 = whisper_full_get_segment_t1(ctx, i);
-
-            start_obj(nullptr);
-                times_o(t0, t1, false);
-                value_s("text", text, false);
-
-                start_arr("tokens");
-                const int n = whisper_full_n_tokens(ctx, i);
-                for (int j = 0; j < n; ++j) {
-                    auto token = whisper_full_get_token_data(ctx, i, j);
-                    start_obj(nullptr);
-                        value_s("text", whisper_token_to_str(ctx, token.id), false);
-                        if(token.t0 > -1 && token.t1 > -1) {
-                            // If we have per-token timestamps, write them out
-                            times_o(token.t0, token.t1, false);
-                        }
-                        value_i("id", token.id, false);
-                        value_f("p", token.p, false);
-                        value_f("t_dtw", token.t_dtw, true);
-                    end_obj(j == (n - 1));
-                }
-                end_arr(true);
-
-            end_obj(i == (n_segments - 1));
-        }
-
-        end_arr(true);
-    end_obj(true);
-
-    if (final) {
-        printf("remotion_final:%s\n", output.c_str());
-    } else {
-        printf("remotion_update:%s\n", output.c_str());
-    }
-
-    return true;
-}
-
-
-void whisper_print_segment_callback(struct whisper_context * ctx, struct whisper_state * /*state*/, int n_new, void * user_data) {
-    const int n_segments = whisper_full_n_segments(ctx);
-    const int s0 = n_segments - n_new;
-
-    if (s0 == 0) {
-        printf("\n");
-    }
-
-    output_json(ctx, false, s0, n_segments);
-}
-
-// Define the progress callback function
-void progress_callback(struct whisper_context * ctx, struct whisper_state * state, int progress, void * user_data) {    
-    printf("remotion_progress:%d%%\n", progress);
-}
-
-std::vector<struct whisper_context *> g_contexts(1, nullptr);
-
-
-EMSCRIPTEN_BINDINGS(whisper) {
-    emscripten::function("full_default", emscripten::optional_override([](const std::string & path_model, const emscripten::val & audio, const std::string & model, const std::string & lang, int nthreads, bool translate) {
-        if (g_contexts[0] != nullptr) {
-            printf("remotion_busy:\n");
-            return 0;
-        }
-
-        g_contexts[0] = whisper_init_from_file_with_params(path_model.c_str(), whisper_context_default_params());
-
-        struct whisper_full_params params = whisper_full_default_params(whisper_sampling_strategy::WHISPER_SAMPLING_GREEDY);
-
-        std::vector<float> pcmf32;
-
-        params.print_realtime   = false;
-        params.new_segment_callback = whisper_print_segment_callback;
-        params.print_progress   = false;
-        params.print_timestamps = false;
-        params.print_special    = false;
-        params.translate        = translate;
-        params.token_timestamps = true;
-        params.language         = lang.c_str(); // Convert std::string to const char*
-        params.n_threads        = std::min(nthreads, std::min(16, mpow2(std::thread::hardware_concurrency())));
-        params.offset_ms        = 0;
-        params.progress_callback = progress_callback; // Assigning the callback
-
-        const int n = audio["length"].as<int>();
-
-        emscripten::val heap = emscripten::val::module_property("HEAPU8");
-        emscripten::val memory = heap["buffer"];
-
-        pcmf32.resize(n);
-
-        emscripten::val memoryView = audio["constructor"].new_(memory, reinterpret_cast<uintptr_t>(pcmf32.data()), n);
-        memoryView.call<void>("set", audio);
-
-        // Print system information
-        {
-            printf("system_info: n_threads = %d / %d | %s\n",
-                   params.n_threads, std::thread::hardware_concurrency(), whisper_print_system_info());
-
-            printf("%s: processing %d samples, %.1f sec, %d threads, lang = %s, task = %s ...\n",
-                   __func__, int(pcmf32.size()), float(pcmf32.size()) / WHISPER_SAMPLE_RATE,
-                   params.n_threads,
-                   params.language,
-                   params.translate ? "translate" : "transcribe");
-
-            printf("\n");
-        }
-
-
-        // Run the worker
-        {
-            g_worker = std::thread([params, pcm = std::move(pcmf32)]() {
-                whisper_reset_timings(g_contexts[0]);
-                whisper_full(g_contexts[0], params, pcm.data(), pcm.size());
-                const int n_segments = whisper_full_n_segments(g_contexts[0]);
-                output_json(g_contexts[0], true, 0, n_segments);
-                whisper_free(g_contexts[0]);
-                g_contexts[0] = nullptr;
-            });
-        }
-
-        return 0;
-    }));
-}

package/eslint.config.mjs

@@ -1,5 +0,0 @@
-import {remotionFlatConfig} from '@remotion/eslint-config-internal';
-
-const config = remotionFlatConfig({react: false});
-
-export default [config];

package/src/can-use-whisper-web.ts

@@ -1,103 +0,0 @@
-import type {WhisperWebModel} from './constants';
-import {SIZES} from './constants';
-
-export enum WhisperWebUnsupportedReason {
-	WindowUndefined = 'window-undefined',
-	IndexedDbUnavailable = 'indexed-db-unavailable',
-	NavigatorStorageUnavailable = 'navigator-storage-unavailable',
-	StorageEstimationApiUnavailable = 'storage-estimation-api-unavailable',
-	QuotaUndefined = 'quota-undefined',
-	UsageUndefined = 'usage-undefined',
-	NotEnoughSpace = 'not-enough-space',
-	ErrorEstimatingStorage = 'error-estimating-storage',
-	NotCrossOriginIsolated = 'not-cross-origin-isolated',
-}
-
-export interface CanUseWhisperWebResult {
-	supported: boolean;
-	reason?: WhisperWebUnsupportedReason;
-	detailedReason?: string;
-}
-
-export const canUseWhisperWeb = async (
-	model: WhisperWebModel,
-): Promise<CanUseWhisperWebResult> => {
-	if (typeof window === 'undefined') {
-		return {
-			supported: false,
-			reason: WhisperWebUnsupportedReason.WindowUndefined,
-			detailedReason:
-				'`window` is not defined. This module can only be used in a browser environment.',
-		};
-	}
-
-	if (!window.crossOriginIsolated) {
-		return {
-			supported: false,
-			reason: WhisperWebUnsupportedReason.NotCrossOriginIsolated,
-			detailedReason:
-				'The document is not cross-origin isolated (window.crossOriginIsolated = false). This prevents the usage of SharedArrayBuffer, which is required by `@remotion/whisper-web`. Make sure the document is served with the HTTP header `Cross-Origin-Opener-Policy: same-origin` and `Cross-Origin-Embedder-Policy: require-corp`: https://remotion.dev/docs/miscellaneous/cross-origin-isolation',
-		};
-	}
-
-	if (!window.indexedDB) {
-		return {
-			supported: false,
-			reason: WhisperWebUnsupportedReason.IndexedDbUnavailable,
-			detailedReason: 'IndexedDB is not available in this environment.',
-		};
-	}
-
-	if (!navigator?.storage || !navigator?.storage.estimate) {
-		return {
-			supported: false,
-			reason: WhisperWebUnsupportedReason.NavigatorStorageUnavailable,
-			detailedReason:
-				'`navigator.storage.estimate()` API is not available in this environment.',
-		};
-	}
-
-	try {
-		const estimate = await navigator.storage.estimate();
-
-		if (estimate.quota === undefined) {
-			return {
-				supported: false,
-				reason: WhisperWebUnsupportedReason.QuotaUndefined,
-				detailedReason:
-					'navigator.storage.estimate() API returned undefined quota.',
-			};
-		}
-
-		if (estimate.usage === undefined) {
-			return {
-				supported: false,
-				reason: WhisperWebUnsupportedReason.UsageUndefined,
-				detailedReason:
-					'navigator.storage.estimate() API returned undefined usage.',
-			};
-		}
-
-		const remaining = estimate.quota - estimate.usage;
-		const modelSize = SIZES[model];
-
-		if (remaining < modelSize) {
-			return {
-				supported: false,
-				reason: WhisperWebUnsupportedReason.NotEnoughSpace,
-				detailedReason: `Not enough space to download the model. Required: ${modelSize} bytes, Available: ${remaining} bytes.`,
-			};
-		}
-	} catch (error) {
-		const errorMessage = error instanceof Error ? error.message : String(error);
-		return {
-			supported: false,
-			reason: WhisperWebUnsupportedReason.ErrorEstimatingStorage,
-			detailedReason: `Error estimating storage: ${errorMessage}`,
-		};
-	}
-
-	return {
-		supported: true,
-	};
-};