address comments

- .github/workflows/swift-format.yml

  2. Documentation/EspeakFramework.md
  - Reorganized documentation to clarify that both macOS and iOS use the same primary flow (packaged bundle first)
  - Separated fallback behavior - only macOS falls back to downloading zip if bundle missing
  - Removed confusion about platform differences in bundling approach

  3. sources/FluidAudio/Resources/.gitkeep & espeak-ng/.gitkeep ❌ DELETED
  - Removed placeholder files (no longer needed if committing actual bundle)

  4. Sources/FluidAudio/TextToSpeech/Kokoro/Assets/Lexicon/EspeakG2P.swift
  - Added configurable espeak voice parameter (defaults to en-us)
  - Tracks current voice and only calls espeak_SetVoiceByName() when it changes
  - Enables multi-language support (British English, Spanish, French, etc.)

  5. Sources/FluidAudio/TextToSpeech/Kokoro/Assets/TtsResourceDownloader.swift
  - Removed PyTorch .pt file download  logic (58 lines removed)
  - Simplified voice embedding download to only fetch usable JSON files
  - Better error messages when embeddings  unavailable

  6.  Sources/FluidAudio/TextToSpeech/Kokoro/Pipeline/Preprocess/KokoroChunker.swift
  - Removed #if canImport(ESpeakNG)  checks (96 lines removed)
  - Removed #available checks for macOS 10.14/iOS 12.0 (unnecessary with current deployment targets)
  - Removed NSLinguisticTagger fallback  code (always uses NaturalLanguage now)
  - Cleaner code without redundant platform checks

Author: Alex-Wengg

.github/workflows/swift-format.yml

@@ -23,10 +23,9 @@ jobs:
       - name: Check formatting
         run: |
           # Create a copy to check if formatting would change files
-          rm -rf Sources.bak Tests.bak Examples.bak
-          rsync -a --safe-links --copy-links Sources/ Sources.bak || true
-          rsync -a --safe-links --copy-links Tests/ Tests.bak || true
-          rsync -a --safe-links --copy-links Examples/ Examples.bak || true
+          cp -r Sources Sources.bak
+          cp -r Tests Tests.bak
+          cp -r Examples Examples.bak || true
           
           # Format in place
           swift format --in-place --recursive --configuration .swift-format Sources/ Tests/ Examples/ || true
@@ -50,4 +49,4 @@ jobs:
             exit 1
           fi
           
-          echo "✅ All files are properly formatted"
+          echo "✅ All files are properly formatted"

Documentation/EspeakFramework.md

@@ -2,15 +2,15 @@
 
 FluidAudio bundles the eSpeak-NG phoneme resources so Kokoro can fall back to G2P lookups when the US lexicons don’t contain a word. The Core ML pipeline expects the resources under `Resources/espeak-ng/espeak-ng-data.bundle` with the canonical `voices/` directory inside.
 
-## macOS (and desktop) builds
-- `TtsResourceDownloader.ensureEspeakDataBundle` first stages the packaged `espeak-ng-data.bundle` directly from the SwiftPM resources.
-- If the packaged copy is removed, macOS falls back to downloading `espeak-ng.zip` and extracts it with `/usr/bin/unzip` into `~/.cache/fluidaudio/Models/kokoro/Resources/`.
-- The `voices/` directory is validated after extraction; if it’s missing we raise `TTSError.downloadFailed`.
+## All Platforms (Primary Flow)
+- `TtsResourceDownloader.ensureEspeakDataBundle` first attempts to stage the packaged `espeak-ng-data.bundle` from SwiftPM resources (`Sources/FluidAudio/Resources/espeak-ng/`).
+- The bundle is copied to `~/.cache/fluidaudio/Models/kokoro/Resources/espeak-ng/`.
+- The `voices/` directory is validated after staging; if missing, `TTSError.downloadFailed` is raised.
 
-## iOS / tvOS / watchOS
-- The Swift package now looks for a pre-packaged `espeak-ng-data.bundle` under `Sources/FluidAudio/Resources/espeak-ng/` and stages it into the cache on first use.
-- If the bundle is missing, we surface `TTSError.downloadFailed`; iOS builds no longer attempt to shell out or download the ZIP on-device.
-- Seed the packaged bundle (or pre-populate the on-device cache) before running TTS on these platforms.
+## Fallback Behavior (macOS Only)
+- If the packaged bundle is unavailable, **macOS only** falls back to downloading `espeak-ng.zip` from HuggingFace and extracting it with `/usr/bin/unzip`.
+- **iOS/tvOS/watchOS** do not support fallback downloads and will throw `TTSError.downloadFailed` if the packaged bundle is missing.
+- For mobile platforms, ensure the packaged bundle is present in the Swift package resources before building.
 
 ## Best practices
 - Keep the `espeak-ng-data.bundle` (packaged copy) and the optional `espeak-ng.zip` fallback in sync with any updates to the Kokoro phoneme mapper.

Sources/FluidAudio/Resources/.gitkeep

@@ -17,9 +17,9 @@ final class EspeakG2P {
         true
     }
 
-    func phonemize(word: String) -> [String]? {
+    func phonemize(word: String, espeakVoice: String = "en-us") -> [String]? {
         return queue.sync {
-            guard initializeIfNeeded() else { return nil }
+            guard initializeIfNeeded(espeakVoice: espeakVoice) else { return nil }
             return word.withCString { cstr -> [String]? in
                 var raw: UnsafeRawPointer? = UnsafeRawPointer(cstr)
                 let modeIPA = Int32(espeakPHONEMES_IPA)
@@ -39,8 +39,16 @@ final class EspeakG2P {
         }
     }
 
-    private func initializeIfNeeded() -> Bool {
-        if initialized { return true }
+    private var currentVoice: String = "en-us"
+
+    private func initializeIfNeeded(espeakVoice: String = "en-us") -> Bool {
+        if initialized {
+            if espeakVoice != currentVoice {
+                _ = espeakVoice.withCString { espeak_SetVoiceByName($0) }
+                currentVoice = espeakVoice
+            }
+            return true
+        }
 
         guard let base = try? TtsModels.cacheDirectoryURL() else {
             logger.warning("Unable to resolve TTS cache directory; disabling eSpeak G2P")
@@ -64,7 +72,8 @@ final class EspeakG2P {
             logger.error("eSpeak NG initialization failed (rc=\(rc))")
             return false
         }
-        _ = "en-us".withCString { espeak_SetVoiceByName($0) }
+        _ = espeakVoice.withCString { espeak_SetVoiceByName($0) }
+        currentVoice = espeakVoice
         initialized = true
         return true
     }

Sources/FluidAudio/Resources/espeak-ng/.gitkeep

@@ -46,59 +46,23 @@ public enum TtsResourceDownloader {
 
     /// Download a voice embedding JSON file from HuggingFace
     public static func downloadVoiceEmbedding(voice: String) async throws -> Data {
-        // Try to download pre-converted JSON first
         let jsonURL = "\(kokoroBaseURL)/voices/\(voice).json"
 
-        if let url = URL(string: jsonURL) {
-            do {
-                let data = try await AssetDownloader.fetchData(
-                    from: url,
-                    description: "\(voice) voice embedding JSON",
-                    logger: logger
-                )
-                logger.info("Downloaded voice embedding JSON for \(voice)")
-                return data
-            } catch {
-                logger.warning("Could not download \(voice).json: \(error.localizedDescription)")
-            }
+        guard let url = URL(string: jsonURL) else {
+            throw TTSError.modelNotFound("Invalid URL for voice embedding: \(voice)")
         }
 
-        var downloadedPtPath: String?
-
-        // Download the .pt file for future conversion
-        let ptURL = "\(kokoroBaseURL)/voices/\(voice).pt"
-        if let url = URL(string: ptURL) {
-            do {
-                let ptData = try await AssetDownloader.fetchData(
-                    from: url,
-                    description: "\(voice) voice embedding (.pt)",
-                    logger: logger
-                )
-
-                let cacheDir = try TtsModels.cacheDirectoryURL()
-                let voicesDir = cacheDir.appendingPathComponent("Models/kokoro/voices")
-                try FileManager.default.createDirectory(at: voicesDir, withIntermediateDirectories: true)
-
-                let ptFileURL = voicesDir.appendingPathComponent("\(voice).pt")
-                try ptData.write(to: ptFileURL, options: [.atomic])
-                downloadedPtPath = ptFileURL.path
-                logger.info(
-                    "Downloaded voice embedding .pt file for \(voice) (\(ptData.count) bytes)")
-                logger.notice(
-                    "Run 'python3 extract_voice_embeddings.py' to convert \(voice).pt to JSON format"
-                )
-            } catch {
-                logger.warning("Could not download \(voice).pt: \(error.localizedDescription)")
-            }
-        }
-
-        if let path = downloadedPtPath {
-            throw TTSError.processingFailed(
-                "Voice embedding JSON unavailable for \(voice). Downloaded .pt to \(path); run 'python3 extract_voice_embeddings.py' to convert it."
+        do {
+            let data = try await AssetDownloader.fetchData(
+                from: url,
+                description: "\(voice) voice embedding JSON",
+                logger: logger
             )
+            logger.info("Downloaded voice embedding JSON for \(voice)")
+            return data
+        } catch {
+            throw TTSError.modelNotFound("Voice embedding JSON unavailable for \(voice): \(error.localizedDescription)")
         }
-
-        throw TTSError.modelNotFound("Voice embedding JSON for \(voice)")
     }
 
     /// Ensure a voice embedding is available in cache

Sources/FluidAudio/TextToSpeech/Kokoro/Assets/Lexicon/EspeakG2P.swift

@@ -385,18 +385,12 @@ enum KokoroChunker {
             phonemes = lexicon[normalized]
         }
 
-        #if canImport(ESpeakNG)
-        if phonemes == nil {
-            if #available(macOS 13.0, iOS 16.0, *) {
-                if let ipa = EspeakG2P.shared.phonemize(word: normalized) {
-                    let mapped = PhonemeMapper.mapIPA(ipa, allowed: allowed)
-                    if !mapped.isEmpty {
-                        phonemes = mapped
-                    }
-                }
+        if phonemes == nil, let ipa = EspeakG2P.shared.phonemize(word: normalized) {
+            let mapped = PhonemeMapper.mapIPA(ipa, allowed: allowed)
+            if !mapped.isEmpty {
+                phonemes = mapped
             }
         }
-        #endif
 
         if phonemes == nil,
             let spelledTokens = spelledOutTokens(for: normalized),
@@ -408,18 +402,12 @@ enum KokoroChunker {
             for spelled in spelledTokens {
                 var segment = lexicon[spelled]
 
-                #if canImport(ESpeakNG)
-                if segment == nil {
-                    if #available(macOS 13.0, iOS 16.0, *) {
-                        if let ipa = EspeakG2P.shared.phonemize(word: spelled) {
-                            let mapped = PhonemeMapper.mapIPA(ipa, allowed: allowed)
-                            if !mapped.isEmpty {
-                                segment = mapped
-                            }
-                        }
+                if segment == nil, let ipa = EspeakG2P.shared.phonemize(word: spelled) {
+                    let mapped = PhonemeMapper.mapIPA(ipa, allowed: allowed)
+                    if !mapped.isEmpty {
+                        segment = mapped
                     }
                 }
-                #endif
 
                 if segment == nil, let fallback = letterPronunciations[spelled] {
                     let filtered = fallback.filter { allowed.contains($0) }
@@ -590,60 +578,18 @@ enum KokoroChunker {
         let breakCharacters = CharacterSet(charactersIn: ",;:")
         let separatorTokens = [": ", "; ", ", "]
 
-        #if canImport(NaturalLanguage)
-        if #available(macOS 10.14, iOS 12.0, *) {
-            let tagger = NLTagger(tagSchemes: [.lexicalClass])
-            tagger.string = text
-            tagger.enumerateTags(
-                in: text.startIndex..<text.endIndex,
-                unit: .word,
-                scheme: .lexicalClass,
-                options: []
-            ) { tag, range in
-                guard tag == .punctuation else { return true }
-                let token = text[range]
-                if token.unicodeScalars.contains(where: { breakCharacters.contains($0) }) {
-                    var endIndex = range.upperBound
-                    for separator in separatorTokens where text[endIndex...].hasPrefix(separator) {
-                        endIndex = text.index(endIndex, offsetBy: separator.count)
-                        break
-                    }
-                    let segment = text[currentStart..<endIndex]
-                    let trimmed = segment.trimmingCharacters(in: .whitespacesAndNewlines)
-                    if !trimmed.isEmpty {
-                        segments.append(trimmed)
-                    }
-                    currentStart = endIndex
-                }
-                return true
-            }
-        } else {
-            let tagger = NSLinguisticTagger(tagSchemes: [.lexicalClass], options: 0)
-            tagger.string = text
-            let nsRange = NSRange(location: 0, length: (text as NSString).length)
-            tagger.enumerateTags(in: nsRange, unit: .word, scheme: .lexicalClass, options: []) {
-                tag, range, _ in
-                guard tag == .punctuation, let tokenRange = Range(range, in: text) else { return }
-                let token = text[tokenRange]
-                if token.unicodeScalars.contains(where: { breakCharacters.contains($0) }) {
-                    var endIndex = tokenRange.upperBound
-                    for separator in separatorTokens where text[endIndex...].hasPrefix(separator) {
-                        endIndex = text.index(endIndex, offsetBy: separator.count)
-                        break
-                    }
-                    let segment = text[currentStart..<endIndex]
-                    let trimmed = segment.trimmingCharacters(in: .whitespacesAndNewlines)
-                    if !trimmed.isEmpty {
-                        segments.append(trimmed)
-                    }
-                    currentStart = endIndex
-                }
-            }
-        }
-        #else
-        for (offset, character) in text.enumerated() {
-            if let scalar = character.unicodeScalars.first, breakCharacters.contains(scalar) {
-                var endIndex = text.index(text.startIndex, offsetBy: offset + 1)
+        let tagger = NLTagger(tagSchemes: [.lexicalClass])
+        tagger.string = text
+        tagger.enumerateTags(
+            in: text.startIndex..<text.endIndex,
+            unit: .word,
+            scheme: .lexicalClass,
+            options: []
+        ) { tag, range in
+            guard tag == .punctuation else { return true }
+            let token = text[range]
+            if token.unicodeScalars.contains(where: { breakCharacters.contains($0) }) {
+                var endIndex = range.upperBound
                 for separator in separatorTokens where text[endIndex...].hasPrefix(separator) {
                     endIndex = text.index(endIndex, offsetBy: separator.count)
                     break
@@ -655,8 +601,8 @@ enum KokoroChunker {
                 }
                 currentStart = endIndex
             }
+            return true
         }
-        #endif
 
         if currentStart < text.endIndex {
             let tail = text[currentStart..<text.endIndex]