Skip to content

[iOS] Add preset X-ASR flow to Qwen3.5 chat #4767

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
ring-zl wants to merge 2 commits into
base: feature/ios-app-dev
Choose a base branch
from
Open

[iOS] Add preset X-ASR flow to Qwen3.5 chat #4767

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
3f97e44
[iOS:Bugfix] Drop unavailable cmark binary target
Aug 13, 2026
d7d1175
[iOS:Feature] Add preset X-ASR to Qwen chat
Aug 13, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 6 additions & 0 deletions apps/iOS/MNNLLMChat/.gitignore
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,10 @@
MNN

# Preset-ASR assets are provisioned separately and must not enter Git.
libsherpa-mnn.a
MNNLLMiOS/LocalModel/preset_audio/
MNNLLMiOS/LocalModel/xasr-mnn-int8/
MNNLLMiOS/LocalModel/Qwen3.5-2B-MNN/
.DS_Store
MNNLLMiOS/LocalModel/Qwen3.5-2B/
_model_holding/
Expand Down
41 changes: 36 additions & 5 deletions ...OS/MNNLLMChat/LocalPackages/Chat/Sources/ExyteChat/ChatView/MessageView/MessageView.swift
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -155,7 +155,7 @@ struct MessageView: View {
attachmentsView(message)
}

if !message.text.isEmpty {
if !message.text.isEmpty, message.recording == nil {
textWithTimeView(message)
.font(Font(font))

Expand All @@ -175,11 +175,17 @@ struct MessageView: View {
}

if let recording = message.recording {
VStack(alignment: .trailing, spacing: 8) {
VStack(alignment: .leading, spacing: 10) {
recordingView(recording)
messageTimeView()
.padding(.bottom, 8)
.padding(.trailing, 12)
if !message.text.isEmpty {
audioEvidenceView(message.text)
}
HStack {
Spacer()
messageTimeView()
}
.padding(.bottom, 8)
.padding(.horizontal, MessageView.horizontalTextPadding)
}
}
}
Expand Down Expand Up @@ -350,6 +356,31 @@ struct MessageView: View {
.padding(.top, 8)
}

@ViewBuilder
func audioEvidenceView(_ text: String) -> some View {
let lines = text.split(separator: "\n", maxSplits: 1, omittingEmptySubsequences: true)
VStack(alignment: .leading, spacing: 4) {
if let transcript = lines.first {
Text(String(transcript))
.font(.caption)
.foregroundColor(message.user.isCurrentUser
? theme.colors.messageMyText.opacity(0.86)
: theme.colors.messageFriendText.opacity(0.86))
.fixedSize(horizontal: false, vertical: true)
}
if lines.count > 1 {
Text(String(lines[1]))
.font(.caption2.monospacedDigit())
.foregroundColor(message.user.isCurrentUser
? theme.colors.messageMyText.opacity(0.62)
: theme.colors.messageFriendText.opacity(0.62))
.fixedSize(horizontal: false, vertical: true)
}
}
.padding(.horizontal, MessageView.horizontalTextPadding)
.accessibilityElement(children: .combine)
}

func messageTimeView(needsCapsule: Bool = false) -> some View {
Group {
if showMessageTimeView {
Expand Down
32 changes: 17 additions & 15 deletions ...S/MNNLLMChat/LocalPackages/Chat/Sources/ExyteChat/ChatView/Recording/RecordWaveform.swift
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,8 +9,6 @@ import SwiftUI

struct RecordWaveformWithButtons: View {

@Environment(\.chatTheme) private var theme

@StateObject var recordPlayer = RecordingPlayer()

// 160 is screen left-padding/right-padding and playButton's width.
Expand All @@ -30,21 +28,25 @@ struct RecordWaveformWithButtons: View {

var body: some View {
HStack(spacing: 12) {
Group {
if recordPlayer.playing {
theme.images.message.pauseAudio
.renderingMode(.template)
} else {
theme.images.message.playAudio
.renderingMode(.template)
}
}
.foregroundColor(colorButton)
.viewSize(40)
.circleBackground(colorButtonBg)
.onTapGesture {
Button {
recordPlayer.togglePlay(recording)
} label: {
Image(systemName: recordPlayer.playing ? "pause.fill" : "play.fill")
.font(.system(size: 17, weight: .semibold))
.foregroundColor(colorButton)
.offset(x: recordPlayer.playing ? 0 : 1)
.frame(width: 44, height: 44)
.background(Circle().fill(colorButtonBg))
.overlay {
Circle()
.stroke(colorButton.opacity(0.18), lineWidth: 1)
}
.shadow(color: Color.black.opacity(0.10), radius: 2, y: 1)
}
.buttonStyle(.plain)
.contentShape(Circle())
.accessibilityLabel(recordPlayer.playing ? "暂停音频" : "播放音频")
.accessibilityHint("播放或暂停样例音频")

VStack(alignment: .leading, spacing: 5) {
RecordWaveformPlaying(samples: recording.waveformSamples, progress: recordPlayer.progress, color: colorWaveform, addExtraDots: false) { progress in
Expand Down
13 changes: 0 additions & 13 deletions apps/iOS/MNNLLMChat/LocalPackages/swift-cmark/Package.swift
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,9 +26,6 @@ let package = Package(
.library(
name: "cmark-gfm-extensions",
targets: ["cmark-gfm-extensions"]),
.executable(
name: "cmark-gfm-bin",
targets: ["cmark-gfm-bin"]),
.executable(name: "api_test",
targets: ["api_test"])
],
Expand All @@ -54,16 +51,6 @@ let package = Package(
],
cSettings: cSettings
),
.target(name: "cmark-gfm-bin",
dependencies: [
"cmark-gfm",
"cmark-gfm-extensions",
],
path: "bin",
sources: [
"main.c",
]
),
.target(name: "api_test",
dependencies: [
"cmark-gfm",
Expand Down
24 changes: 24 additions & 0 deletions apps/iOS/MNNLLMChat/MNNLLMiOS.xcodeproj/project.pbxproj
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,6 +27,9 @@
3ECE19C72E9CE1BD001A4CDD /* TextFileActivityItemSource.swift in MNNChat */ = {isa = PBXBuildFile; fileRef = 3ECE19C62E9CE1BD001A4CDD /* TextFileActivityItemSource.swift */; };
3ECE19C92E9CE1F0001A4CDD /* BatchFileTestService.swift in MNNChat */ = {isa = PBXBuildFile; fileRef = 3ECE19C82E9CE1F0001A4CDD /* BatchFileTestService.swift */; };
3E000000000000000000A102 /* PresetPrompts.swift in MNNChat */ = {isa = PBXBuildFile; fileRef = 3E000000000000000000A101 /* PresetPrompts.swift */; };
5A000000000000000000A107 /* PresetZipformerASRService.swift in MNNChat */ = {isa = PBXBuildFile; fileRef = 5A000000000000000000A101 /* PresetZipformerASRService.swift */; };
5A000000000000000000A104 /* PresetZipformerASRWrapper.mm in MNNChat */ = {isa = PBXBuildFile; fileRef = 5A000000000000000000A103 /* PresetZipformerASRWrapper.mm */; };
5A000000000000000000A106 /* libsherpa-mnn.a in Frameworks */ = {isa = PBXBuildFile; fileRef = 5A000000000000000000A105 /* libsherpa-mnn.a */; };
3EEAB30B2E8BD5550039DF50 /* BatchFileTestViewModel.swift in MNNChat */ = {isa = PBXBuildFile; fileRef = 3EEAB30A2E8BD5550039DF50 /* BatchFileTestViewModel.swift */; };
4230C2D22E7D4EB5001169DF /* Preview Assets.xcassets in Resources */ = {isa = PBXBuildFile; fileRef = 4230C2A22E7D4EB5001169DF /* Preview Assets.xcassets */; };
4230C2DB2E7D4EB5001169DF /* mock.json in Resources */ = {isa = PBXBuildFile; fileRef = 4230C2AD2E7D4EB5001169DF /* mock.json */; };
Expand Down Expand Up @@ -156,6 +159,10 @@
3ECE19C62E9CE1BD001A4CDD /* TextFileActivityItemSource.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = TextFileActivityItemSource.swift; sourceTree = "<group>"; };
3ECE19C82E9CE1F0001A4CDD /* BatchFileTestService.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = BatchFileTestService.swift; sourceTree = "<group>"; };
3E000000000000000000A101 /* PresetPrompts.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = PresetPrompts.swift; sourceTree = "<group>"; };
5A000000000000000000A101 /* PresetZipformerASRService.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = PresetZipformerASRService.swift; sourceTree = "<group>"; };
5A000000000000000000A102 /* PresetZipformerASRWrapper.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; path = PresetZipformerASRWrapper.h; sourceTree = "<group>"; };
5A000000000000000000A103 /* PresetZipformerASRWrapper.mm */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.cpp.objcpp; path = PresetZipformerASRWrapper.mm; sourceTree = "<group>"; };
5A000000000000000000A105 /* libsherpa-mnn.a */ = {isa = PBXFileReference; lastKnownFileType = archive.ar; path = libsherpa-mnn.a; sourceTree = "<group>"; };
3EEAB30A2E8BD5550039DF50 /* BatchFileTestViewModel.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = BatchFileTestViewModel.swift; sourceTree = "<group>"; };
4230C22C2E7D4EB5001169DF /* ChatInteractorProtocol.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = ChatInteractorProtocol.swift; sourceTree = "<group>"; };
4230C22D2E7D4EB5001169DF /* LLMChatInteractor.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = LLMChatInteractor.swift; sourceTree = "<group>"; };
Expand Down Expand Up @@ -284,6 +291,7 @@
isa = PBXFrameworksBuildPhase;
buildActionMask = 2147483647;
files = (
5A000000000000000000A106 /* libsherpa-mnn.a in Frameworks */,
3ECAC6062F3C62570003C496 /* ExyteChat in Frameworks */,
3E2F97302E89158600E61F35 /* ExyteChat in Frameworks */,
3E2F972F2E89158600E61F35 /* ExyteChat in Frameworks */,
Expand All @@ -303,6 +311,7 @@
isa = PBXGroup;
children = (
3E440F082EE2DEDE00BB15FF /* AudioPlaybackManager.swift */,
5A000000000000000000A101 /* PresetZipformerASRService.swift */,
);
path = Audio;
sourceTree = "<group>";
Expand Down Expand Up @@ -332,6 +341,7 @@
isa = PBXGroup;
children = (
3E301C682D5C84730045E5E1 /* MNN.framework */,
5A000000000000000000A105 /* libsherpa-mnn.a */,
);
name = Frameworks;
sourceTree = "<group>";
Expand Down Expand Up @@ -458,6 +468,8 @@
4230C2522E7D4EB5001169DF /* InferenceEngine */ = {
isa = PBXGroup;
children = (
5A000000000000000000A102 /* PresetZipformerASRWrapper.h */,
5A000000000000000000A103 /* PresetZipformerASRWrapper.mm */,
3E2B293A2F1F7C28006DE361 /* SanaDiffusionSession.h */,
3E2B293B2F1F7C28006DE361 /* SanaDiffusionSession.mm */,
4230C24E2E7D4EB5001169DF /* DiffusionSession.h */,
Expand Down Expand Up @@ -890,6 +902,8 @@
isa = PBXSourcesBuildPhase;
buildActionMask = 2147483647;
files = (
5A000000000000000000A107 /* PresetZipformerASRService.swift in MNNChat */,
5A000000000000000000A104 /* PresetZipformerASRWrapper.mm in MNNChat */,
4230C2DF2E7D4EB5001169DF /* ChatInteractorProtocol.swift in MNNChat */,
4230C2E02E7D4EB5001169DF /* LLMChatInteractor.swift in MNNChat */,
4230C2E12E7D4EB5001169DF /* ThinkResultProcessor.swift in MNNChat */,
Expand Down Expand Up @@ -1175,6 +1189,11 @@
"INFOPLIST_KEY_UIStatusBarStyle[sdk=iphonesimulator*]" = UIStatusBarStyleDefault;
INFOPLIST_KEY_UISupportedInterfaceOrientations = "UIInterfaceOrientationLandscapeLeft UIInterfaceOrientationLandscapeRight UIInterfaceOrientationPortrait UIInterfaceOrientationPortraitUpsideDown";
IPHONEOS_DEPLOYMENT_TARGET = 17;
HEADER_SEARCH_PATHS = "$(SRCROOT)/../../frameworks/sherpa-mnn";
LIBRARY_SEARCH_PATHS = (
"$(inherited)",
"$(PROJECT_DIR)",
);
LD_RUNPATH_SEARCH_PATHS = "@executable_path/Frameworks";
"LD_RUNPATH_SEARCH_PATHS[sdk=macosx*]" = "@executable_path/../Frameworks";
MACOSX_DEPLOYMENT_TARGET = 14;
Expand Down Expand Up @@ -1232,6 +1251,11 @@
"INFOPLIST_KEY_UIStatusBarStyle[sdk=iphonesimulator*]" = UIStatusBarStyleDefault;
INFOPLIST_KEY_UISupportedInterfaceOrientations = "UIInterfaceOrientationLandscapeLeft UIInterfaceOrientationLandscapeRight UIInterfaceOrientationPortrait UIInterfaceOrientationPortraitUpsideDown";
IPHONEOS_DEPLOYMENT_TARGET = 17;
HEADER_SEARCH_PATHS = "$(SRCROOT)/../../frameworks/sherpa-mnn";
LIBRARY_SEARCH_PATHS = (
"$(inherited)",
"$(PROJECT_DIR)",
);
LD_RUNPATH_SEARCH_PATHS = "@executable_path/Frameworks";
"LD_RUNPATH_SEARCH_PATHS[sdk=macosx*]" = "@executable_path/../Frameworks";
MACOSX_DEPLOYMENT_TARGET = 14;
Expand Down
Loading