First Open Beta Release

• recognition accuracy improvement by @shibukazu in #168
• fix to non-block the recording on background by @shibukazu in #183
• set max width by @yasutak in #185
• modified to show ongoing recognition results by @shibukazu in #160
• add Package.resolved by @yasutak in #186
• add missing file by @yasutak in #189
• Add info page by @yasutak in #178
• Activate swipe gesture for side menu closing by @shibukazu in #181
• prooperly added package.resolved to gitignore by @yasutak in #197
• model download feature by @yasutak in #174
• Improvement/splash screen by @Hwari2929 in #212
• audio-sharing feature by @Hwari2929 in #194
• fix view issues by @ooyamatakehisa in #201
• delete developer page by @yasutak in #204
• Add english localization by @yasutak in #221
• update initial image by @yasutak in #233
• update icons by @yasutak in #235
• language and model select pane by @shibukazu in #223
• remove white blank from splash-screen by @shibukazu in #244
• Suppress audio and transcript time mismatch by @shibukazu in #243
• add missing translation by @yasutak in #245
• change model to int4 quantization model by @shibukazu in #246
• replace car with hare by @yasutak in #248
• change non_speech_token_suppression true to false for Japanese by @shibukazu in #252
• fix default model by @shibukazu in #255
• add repetition suppression feature by @shibukazu in #257
• remove focus during recognition by @yasutak in #259

Full Changelog: v0.1.0-beta...v0.2.1-beta

First Beta Release Candidate

First Beta Release Candidate (#253)

* recognition accuracy  improvement (#168)

* update whisper.cpp and activate non_speech_token suppression

* add prompt reset feature

* modify remaining audio concatination

* fix to non-block the recording on background (#183)

* fix to non-block the recording on background

* fix to distinguish inactive and background

* format

* set max width (#185)

* set max width

* apply formatter

* fixes based on comments

* modified to show ongoing recognition results (#160)

* modified to show ongoing recognition results

* changed to pop up alert when recognition is on going

* format

* add Package.resolved (#186)

* add Package.resolved

* add .gitignore itself

* add missing file (#189)

* Add info page (#178)

* add app info menu bar

* show webpages with in-app browser

* format

* add padding

* update terms of service url

* fixes based on commnets

* additional fixes

* specify fonts

* Activate swipe gesture for side menu closing (#181)

* add smooth sidemenu close feature

* fix to activate record button

* format

* prooperly added package.resolved to gitignore (#197)

* model download feature (#174)

* add pull request template (#77)

* add model path

* add whisper model definition

* add download function

* move url list

* add repository functions

* add missing self

* enable itunes file-sharing and file app integration

* initiate whisper model in menu view

* make WhisperRecognizer accept WhisperModel

* resolve compile errors

* return URL type

* fix error for guard statment

* fix error

* replace model name with WhisperModel

* uppercase struct names

* change @AppStorage objects

* make storage-key as String

* add initilizers

* fix type error

* fix assignment errors

* fix display name

* remove unused property

* fix key names

* add missing args

* fix wrong operator

* refactoring audio player (#95)

* fix typos

* remove `playerWrapper` and use raw `AVAudioPlayer`

* format

* remove  state

* add whisperModel to workspace

* add whisperModelRepository to repo

* make whisperModel initializer accept callback function so that recognizer's model loading occurs after the completion of download

* change model name

* update recognizer.whisperModel outside model initialization in order to avoid 'published property cannot be updated outside main thread' error

* load models when they already exist

* set whisperModel in initialization

* explicitly set whisperModel path in initialization

* add code for error handling

* Revert "enable itunes file-sharing and file app integration"

This reverts commit f5d7d2d9a44a8cce242659492d2c7aee2ad1936d.

* apply formatter

* extract downloader

* format

* free whisperContext before load

* format

* fix comment format

* undo change in buildConfiguration

* fix some appearance

* checkmarks for downloads work somehow

* swap wrong icons

* refactor

* show prompt for downloading

* separate model download and model loading

* to avoid annoying confirmation whenever I submit testflight builds

* change icon size

* beatify progress bar design

* change progress bar color

* progress bar working

* reorganize callback

* update value in main thread

* delete model gesture working

* delete model functino complete

* remove cloud icons for tiny models

* fix selected-recognizer-reseted-bug #125

* load models asynchronously

* remove unused suscess

* disable delete while loading

* add clearer var name

* remove unnecesarry Dispatch

* add ggml-tiny.multi.bin

* format

* add new formatting rule to limit max width to 120

* reformat

* update formtter version

* reformat

* change bundle identifier

* add tiny model

* remove inject from dependencies

* add medium model

* remove unnecesarry files

* remove wrongly enalbed UIApplicationSupportsMultipleScenes

* translate english comments into japanese ones

* disanle multiwindows

* Co-authored-by: Takehisa Oyama <[email protected]>

* update file info

* simplify logic

* use callback for initializing recognizer on app start

* remove unused needssubscription

* format

* remove optional flag

* some fixes

* refactoring

* fix bugs

* fix bug

---------

Co-authored-by: Takehisa Oyama <[email protected]>
Co-authored-by: ooyamatakehisa <[email protected]>

* Improvement/splash screen (#212)

* splash-screen improvement

* splash-screen

* rolled back entitlements file

* rollback HomeView

* rollback

* roll back unnecessary

* format

* audio-sharing feature (#194)

* audio-sharing

* formatted

* change button's position

* format

* レビューを反映しました。

* Update ToolBar.swift

* audio bar width restore

* fix view issues (#201)

* re-add resource to make preview work

* remove animation from stop button

* resolve moving red circle animation issue

* remove extra space above navigation bar

* manage preview codes

* set circle height

* fix too big waveform

* change app name

* set `maxAmp` to 0 when starting recording

---------

Co-authored-by: shibukazu <[email protected]>

* delete developer page (#204)

* delete developer page

* remove StreamingRecognition row in project.pbxproj

---------

Co-authored-by: shibukazu <[email protected]>

* Add english localization (#221)

* add english localization

* enable language switching

* add alert for potential data loss

* add message for macOS

* add missing translation

* hide switch item when there is only one display language installed on device

* fixes based on comments

* additional fixes

* remove language switch feature

* add missing translation

* change display name to VoiScribe

* update initial image (#233)

* update icons (#235)

* language and model select pane (#223)

* delete Package.resolved

* add partial view to set recognition model and language before start recording

* change to use presets and fix each box size

* enabled user to download model from sheet

* fix design and add model size information

* refactoring

* delete unused files

* fix to share model download state

* fix to share isDownloading state

* removed tiny models and changed default model

* add recommend tag

* fix to prohibit model change when recognition on progress

* fix folder name

* format

* fix for mac catalyst

* fix displayed model size

* add localization

* removed recognition frequency sec item in sidemenu

* remove model load feature from sidemenu

* format

* add license information

* fix design for english environment

* change default recognition language and model based on device setting

* format

* fix to show sheet in stead of partial sheet in iPad

* format

* add defer statement

* format

* fix to init isDownloading value as false when app relaunched

* add error handling

* remove white blank from splash-screen (#244)

* remove white blank from splash-screen

* temporarily change github actions

* fix swift-format version

* suppress hallucination and improved audio-transcript-time matching (#243)

* add missing translation (#245)

* change model to int4 quantization model (#246)

* update whisper library

* change small-en and medium-en to quantized model

* change model megabytes calculation

* format

* replace car with hare (#248)

* Update README.md

* change non_speech_token_suppression true to false for Japanese (#252)

---------

Co-authored-by: shibukazu <[email protected]>
Co-authored-by: Takehisa Oyama <[email protected]>
Co-authored-by: ooyamatakehisa <[email protected]>
Co-authored-by: Hwari2929 <[email protected]>
Co-authored-by: shibukazu <[email protected]>

Initial release

What's Changed

• Recoding button by @ooyamatakehisa in #1
• speech recognition feature by @shibukazu in #2
• add compiler flag by @shibukazu in #3
• add recognition pane and display waveform and recording time by @ooyamatakehisa in #5
• Design/list by @shibukazu in #6
• Add formatter action by @yasutak in #7
• Add Formatter by @yasutak in #8
• save RecognizedSpeech to CoreData by @ooyamatakehisa in #9
• fix committed files by @ooyamatakehisa in #10
• fix data used in preview by @ooyamatakehisa in #11
• fix design and add new mock data by @shibukazu in #12
• improve recognition pane and implement delete feature by @ooyamatakehisa in #13
• sort record descending by @ooyamatakehisa in #14
• implement confirm page by @ooyamatakehisa in #20
• implement audio player by @ooyamatakehisa in #27
• add splash screen by @ooyamatakehisa in #29
• add lodaing icon by @ooyamatakehisa in #30
• volume up by @ooyamatakehisa in #32
• implement automatic scroll by @ooyamatakehisa in #34
• fix strange animation on deletion by @ooyamatakehisa in #35
• サイドメニューの刷新 by @shibukazu in #45
• update recorging pane by @ooyamatakehisa in #51
• add message in RecordDetail when result is empty by @ooyamatakehisa in #53
• enable tapping in entire row by @ooyamatakehisa in #55
• fix RecordList view in ipad by @ooyamatakehisa in #56
• implement initial page by @ooyamatakehisa in #57
• implement share button by @ooyamatakehisa in #60
• Configure SwiftFormat by @yasutak in #19
• ストリーミング認識の実装 by @shibukazu in #73
• 認識画面のデザインを修正 by @shibukazu in #75
• 言語設定のバグを修正 by @shibukazu in #76
• add pull request template by @yasutak in #77
• logger refactoring by @shibukazu in #82
• reactoring streaming asr by @shibukazu in #88
• refactoring audio player by @ooyamatakehisa in #95
• Removed app title in RecordDetails view, but 'Notes' navigation bar t… by @jaeyounkg in #91
• fix isPlaying issue by @ooyamatakehisa in #99
• update icon by @yasutak in #109
• prompting and audio concatenation feature by @shibukazu in #94
• change development team id by @ooyamatakehisa in #129
• fix bug where audio is reset when close the pane by @ooyamatakehisa in #130
• fix bug where recognizing speech cannot be deleted by @ooyamatakehisa in #132
• bug fix for isActives variable by @shibukazu in #134
• copy and edit feature by @ooyamatakehisa in #102
• add recording interruption handler by @shibukazu in #139
• enable to build in iOS15 by @ooyamatakehisa in #148
• execute timer on another thread by @ooyamatakehisa in #147
• enable background playback and recording by @ooyamatakehisa in #141
• fix missing AVFoundation by @ooyamatakehisa in #167

New Contributors

• @yasutak made their first contribution in #7
• @jaeyounkg made their first contribution in #91

Full Changelog: https://github.com/openly-jp/whisper-ios/commits/v0.1.0-beta