iOS SDK

Note:

Supports iOS versions 11.0 and above.

Only activation via internet connection is supported.

1 Integration Steps

Note: If you are interested in trying the on-device recognition feature, please get in touch with our business contact in advance to receive an APPID. Also, please specify the language type you want to try. Contact: voice.contact@dolphin-ai.jp.

Manual import: Drag SpeechEvaluate.framework into your project. Then, under General -> Frameworks, Libraries, and Embedded Content, change the Embed setting for SpeechRecognitionSDK.framework to Embed&Sign.
Ensure the following pods are included: SocketRocket0.6.0, AFNetworking, SSZipArchive2.4.0.
Offline sdk add model file: Place the model file (.zip) into the framework folder.

Add Privacy - Microphone Usage Description to your project's info.plist file to add microphone access permissions.

1.2 Invocation Steps/Sample Code

In the file that requires the recognition function, please adhere to the delegate protocol EvalListener.

//Configuration Main Parameters
SDKParams *params = [[SDKParams alloc] init];
params.appId = @"";
params.appSecret = @"";
params.sample_rate = 16000;
params.format = @"pcm";
params.realtime = NO;
params.langType = @"ja-JP";
params.enable_intermediate_result = YES;
params.enable_punctuation_prediction = YES;
params.max_sentence_silence = 450;
params.enable_words = YES;

/**
@sourcePath Customized paths
@Documents/userPatu/model，The path before the model can be named arbitrarily.
@The vad file must be under the model file, for example：Documents/userPath/model/vad
@The language file must be under the model/asr file, for example：Documents/userPath/model/asr/ja-JP
*/
params.sourcePath = [[NSSearchPathForDirectoriesInDomains(NSDocumentDirectory, NSUserDomainMask, YES) firstObject] stringByAppendingPathComponent:@"/userPath/model"];

1.2.1 Create a Speech Recognition Class and Grant Authorization

Name	Type	Description
SpeechRecognition	SpeechRecognition	Recognition class
params	SDKParams	Parameters and settings

self.params.initType = YES;
self.params.online = NO;
SpeechRecognition *speechManger = [[SpeechRecognition alloc] init];
[speechManger setInitSDK:self params:params];
self.speechManger = speechManger;

1.2.2 Callback Method Description

Name	Type	Description
onStart	String	Callback method when engine connection starts
onResult	String	Callback method when engine returns results
onRealtimeResult	String	Callback method when engine returns intermediate results
onWarning	String	Callback method when engine returns a result warning
onError	String	Callback method when engine returns a result error
onMachineCode	String	Engine returns machine code

- (void) onRealtimeResult: (NSString *) result;

- (void) onResult: (NSString *) result;

- (void) onStart: (NSString *) taskId;

- (void) onStop;

- (void) onGetAudio: (NSData *)data;

- (void) onError: (NSString *)code msg:(NSString*)msg taskId:(nullable NSString*)taskId;

- (void) onWarning: (NSString *)code msg:(NSString*)msg taskId:(nullable NSString*)taskId;

- (void) onMachineCode: (NSString *)MachineCode;

1.2.3 Interface Parameter Description

Parameter	Type	Required	Description	Default Value
lang_type	String	Yes	Language option	Required
format	String	No	Audio encoding format	pcm
sample_rate	Integer	No	Audio sampling rate	16000
enable_intermediate_result	Boolean	No	Whether to return intermediate recognition results	false
enable_punctuation_prediction	Boolean	No	Whether to add punctuation in post-processing	false
enable_inverse_text_normalization	Boolean	No	Whether to perform ITN in post-processing	false
max_sentence_silence	Integer	No	Speech sentence breaking detection threshold. Silence longer than this threshold is considered as a sentence break. The valid parameter range is 200～1200. Unit: Milliseconds	450
enable_words	Boolean	No	Whether to return word information	false
enable_modal_particle_filter	Boolean	No	Whether to enable modal particle filtering	false
hotwords_id	String	No	Hotwords ID	None
hotwords_weight	Float	No	Hotwords weight, the range is [0.1, 1.0]	0.4

1.2.4 Start/Stop Recognition

<1>Start Recognition (Internal Recording by SDK)
    [self.speechManger startRecording];
    
End Recognition
    [self.speechManger stopRecording];
    
<2>File Recognition (Directly pass the path of the audio file, local path)   
    [self.speechManger startRecognitionOralWithWavPath:@"wav audio path"];

<3>Audio Data Recognition (External recording by SDK, or file converted to NSData for recognition)    
- (void)doStart:(FinishBlock)finishBlock;
- (BOOL)doSetData:(NSData *) data isLast:(bool)isLast;
Call Method
[self.speechManger doStart:^(_Bool success) {
            if (success) {
If(Last Segment With Audio) {
[self.speechManger doSetData:data isLast:YES];
}else{
[self.speechManger doSetData:data isLast:NO];
}
}
}];

1.2.5 Load Hotwords

[self.speechManger loadWords];

1.2.6 Destroy on Exit

//Release the current model
[self.speechManger offLineRelese];

//Release all models
[self.speechManger offLineReleseAll];

1.2.7 Cancel the generation of logs

Set the parameters in model/asr/ja-JP/config:
log-level = 6
log-rotate-days=0

2 Status Code Table

Error Code	Error Message	Description	Solution
110100	Unauthorized	Unauthorized	Change the parameter to an authorized capability or contact the business department to add AI capabilities
110101	Invalid Activation Code	Activation key is incorrect	Enter the correct activation key
110102	Model Count Exceeds Limit	The number of models exceeds the limit	Please contact the business department
110103	Authorization Expired	Authorization has expired	Contact the business department to extend the authorization period
110104	Authorization Failed	Authorization failed	Please contact the business department

3 SDK Download & On-Device Model Download

3.1 SDK Download

The download link for the SDK is as follows: iOS SDK

iOS Demo

3.2 On-Device Model Download

Language Type	Language Code	Download Link	Remarks
Japanese - General	ja-JP	Japanese - General	Only Japanese
Japanese - Hotel	ja-JP	Japanese - Hotel	Supports mixed speech of Japanese and English
English	en-US	English
Mandarin	zh-cmn-Hans-CN	Mandarin	Supports mixed speech of Mandarin and English

On this page