Updated the Capturing procotol in the Recording package target to pass captured audio as a file URL instead of data.

This commit is contained in:
2026-07-04 15:06:55 +02:00
parent 2fe240de1a
commit 6a0e7698a5
10 changed files with 116 additions and 119 deletions
@@ -2,7 +2,7 @@ import AVFoundation
/// The recording service that captures real audio from the device's microphone.
///
/// The service records into a temporary `.m4a` file through an `AVAudioRecorder`, and returns the file's contents when the recording stops.
/// The service records into a temporary `.m4a` file through an `AVAudioRecorder`, and returns the file's location when the recording stops.
/// The user's permission to record is requested before a recording starts and, on the platforms that require it, the shared audio session is
/// configured for recording while the capture is in progress.
@MainActor
@@ -73,12 +73,11 @@ public final class AudioCapturing: Capturing {
}
}
/// Stops the recording, deleting the temporary audio file after its contents are read.
/// Stops the recording, handing the temporary audio file over to the caller, which becomes responsible for deleting it.
///
/// - Returns: The audio captured since the recording started.
/// - Throws: ``AudioCapturingError/noOngoingRecording`` when no recording is in progress, or any error thrown
/// while reading the recorded audio file.
public func stop() async throws -> Data {
/// - Returns: The location of the audio file captured since the recording started.
/// - Throws: ``AudioCapturingError/noOngoingRecording`` when no recording is in progress.
public func stop() async throws -> URL {
guard let recorder else {
throw AudioCapturingError.noOngoingRecording
}
@@ -91,11 +90,7 @@ public final class AudioCapturing: Capturing {
try? AVAudioSession.sharedInstance().setActive(false)
#endif
defer {
try? FileManager.default.removeItem(at: Constant.File.url)
}
return try Data(contentsOf: Constant.File.url)
return Constant.File.url
}
}
@@ -3,10 +3,10 @@ import Speech
/// The transcribing service that transcribes recorded audio into text on device.
///
/// The service writes the recorded audio into a temporary `.m4a` file and runs it through a `SpeechAnalyzer` with a `SpeechTranscriber`
/// module, joining the finalized results into the returned transcription. The transcription happens in the locale given at each call or rather in the
/// closest equivalent the transcriber supports. The speech model assets for that locale are downloaded and installed on first use; every
/// transcription after that happens entirely offline.
/// The service runs the given recorded audio file through a `SpeechAnalyzer` with a `SpeechTranscriber` module, joining the finalized
/// results into the returned transcription. The transcription happens in the locale given at each call or rather in the closest equivalent
/// the transcriber supports. The speech model assets for that locale are downloaded and installed on first use; every transcription after
/// that happens entirely offline.
public struct AudioTranscribing: Transcribing {
// MARK: Initializers
@@ -16,22 +16,20 @@ public struct AudioTranscribing: Transcribing {
// MARK: Methods
/// Transcribes the given recorded audio into text, deleting the temporary audio file when the transcription finishes.
/// Transcribes the given recorded audio file into text, deleting the file when the transcription finishes.
///
/// - Parameters:
/// - audio: The recorded audio to transcribe.
/// - audio: The location of the recorded audio file to transcribe.
/// - locale: The locale of the spoken language to transcribe.
/// - Returns: The transcription of the recorded audio.
/// - Throws: ``AudioTranscribingError/localeNotSupported`` when the transcriber supports no equivalent of the given locale,
/// or any error thrown while installing the speech model assets, reading the audio file, or analyzing its contents.
public func callAsFunction(
_ audio: Data,
_ audio: URL,
locale: Locale
) async throws -> Transcription {
try audio.write(to: Constant.File.url)
defer {
try? FileManager.default.removeItem(at: Constant.File.url)
try? FileManager.default.removeItem(at: audio)
}
guard let locale = await SpeechTranscriber.supportedLocale(
@@ -55,7 +53,7 @@ public struct AudioTranscribing: Transcribing {
text += String(result.text.characters)
}
let file = try AVAudioFile(forReading: Constant.File.url)
let file = try AVAudioFile(forReading: audio)
if let lastSampleTime = try await analyzer.analyzeSequence(from: file) {
try await analyzer.finalizeAndFinish(through: lastSampleTime)
@@ -75,14 +73,3 @@ public enum AudioTranscribingError: Error {
/// The transcriber supports no equivalent of the locale the transcription was requested with.
case localeNotSupported
}
// MARK: - Constants
/// The constant values used across the audio transcribing service.
private enum Constant {
/// The file constants.
enum File {
/// The location of the temporary file the recorded audio is written into for the analysis.
static let url = FileManager.default.temporaryDirectory.appending(path: "transcription.m4a")
}
}
@@ -19,9 +19,20 @@ struct DummyCapturing: Capturing {
/// Simulates the stop of a recording.
///
/// - Returns: An empty audio payload.
func stop() async throws -> Data {
.init()
/// - Returns: The location of a dummy audio file, which is never actually created.
func stop() async throws -> URL {
FileManager.default.temporaryDirectory.appending(path: Constant.File.name)
}
}
// MARK: - Constants
/// The constant values used across the recording services.
private enum Constant {
/// The file constants.
enum File {
/// The name of the dummy audio file.
static let name = "dummy.m4a"
}
}
@@ -11,11 +11,11 @@ struct DummyTranscribing: Transcribing {
/// Simulates the transcription of the given recorded audio with a two-second delay.
///
/// - Parameters:
/// - audio: The recorded audio to transcribe.
/// - audio: The location of the recorded audio file to transcribe.
/// - locale: The locale of the spoken language to transcribe.
/// - Returns: A dummy transcription.
func callAsFunction(
_ audio: Data,
_ audio: URL,
locale: Locale
) async throws -> Transcription {
try await Task.sleep(for: Constant.Delay.transcribing)