This document describes the English word repetition speech evaluation technology for children under 12. It supports syllable stress detection, differentiation between British and American accents, and multi-dimensional scoring feedback.-Smart Science and Education Content Platform(SSECP)-阿里云帮助中心

This document describes the speech evaluation technology for word repetition tasks. This technology is designed for developers and is suitable for evaluating the pronunciation of children under 12.

Task description

Evaluates English words and phrases.
Supports scoring based on custom pronunciations, syllable stress detection, and differentiation between British and American accents.
Feedback dimensions include the total word score, stress evaluation, and phoneme scores.

Evaluation request parameters

Field	Type	Optional	Description
coreType	string	Required	en.word_kid.score indicates English word evaluation for children.
refText	string	Required	The reference text for scoring. Note: Specify external pronunciations in two ways: 1. directory(z:d ih - 'r eh k - tr ih): The phonemes follow the internal phoneme specifications of the DPI engine. 2. directory(p:dəˈrektəri): Specifies the pronunciation using the original phonemes. If the original phonemes correspond to multiple standards, the processing priority is IP88 > KK > IP66. For more information, see Text symbol description.
rank	int	Optional	The scoring scale. You can set this value to any integer. The final score is rescaled based on a 100-point scale.
precision	double	Optional	The scoring precision. Valid values are 0.1, 0.5, and 1. The default value is 1. If you set this parameter to a value other than 0.1 or 0.5, the system uses 1.
attachAudioUrl	int	Optional	Specifies whether to return the audio URL in the evaluation result. 0: Does not return the URL. 1: Returns the URL.
typeThres	int	Optional	The scoring strictness. A higher value has less impact on the score. Valid values:1: Strict2: Loose3: Very strict4: Very loose0: Default. This feature is disabled.
accent	string	Optional	Specifies British or American English for scoring. By default, this feature is disabled, and the DPI engine intelligently determines the accent based on the user's pronunciation. Set "accent" to "en" to specify British English evaluation. Set "accent" to "am" to specify American English evaluation.

Parameter example

{
    "coreType":"en.word_kid.score",
    "refText":"egg",
    "rank":100,
    "attachAudioUrl":1,
    "accent":"en"
}

Sample response

{
    "tokenId": "6114e85388bf5b00000a0001",
    "applicationId": "aXXX",            //  The provided appkey
    "audioUrl": "http:\/\/files.cloud.ssapi.cn:8080\/a148\/11ebfb4ecda877958771a148m2611197",  // User audio download URL
    "connect": {   // Internal parameter. For internal use only.
        "param": {
            "app": {
                "timestamp": "1628760147",
                "applicationId": "aXXX",
                "sig": "5f82347d335449cc9aa053a4f2f29820f54bbd2f"
            },
            "sdk": {
                "os": "6.2(9200)1",
                "product": "",
                "source": 7,
                "protocol": 1,
                "os_version": "6.2(9200)",
                "arch": "586",
                "version": 16778752
            }
        },
        "cmd": "connect"
    },
    "params": {
        "app": {
            "timestamp": "1628760147",
            "userId": "ssound_text",
            "sig": "5f82347d335449cc9aa053a4f2f29820f54bbd2f",
            "connect_id": "6114e85388bf5b00000c0002",
            "clientId": "",
            "applicationId": "a148"
        },
        "audio":{                    // Audio format parameters for the scoring request
            "sampleRate":16000,  // Sample rate
            "channel":1,    // Sound channel. 1 indicates mono, and 2 indicates stereo. Evaluation supports only mono.
            "sampleBytes":2,
            "audioType":"ogg"  // Audio type
        },
        "request": {        // Parameter node for the scoring request
            "request_id": "6114e85388bf5b00000b0001",
            "accent": "en",    // This field is returned only if the accent parameter is not configured in the request.
            "rank": 100,                 // The scoring scale. The default value is 100. You can set this to any integer. The final score is rescaled based on a 100-point scale.
            "attachAudioUrl": 1,            // Specifies whether to return the audio URL in the evaluation result. 0: No. 1: Yes.
            "tokenId": "6114e85388bf5b00000a0001",
            "refText": "egg",                       // Text content
            "coreType": "en.word_kid.score"    // Task type
        }
    },
    "recordId": "11ebfb4ecda877958771a148m2611197", // The unique ID of the data in the cloud. Save this to your product database for troubleshooting.
    "refText": "egg",                              // Text content
    "dtLastResponse": "2021-08-12 17:22:27:884",         // Data response timestamp
    "cloud_platform": {
        "origin_audio_length": 7807
    },
    "result": {
        "pron": 98,
        "forceout": 0,
        "precision": 1,
        "systime": 2104,
        "res": "eng.wrd_kid.online.0.2",
        "version": "0.0.80.2021.7.24.00:01:28",
        "info": {
            "tipId": 10004,
            "clip": 0,
            "snr": 30.070004,
            "volume": 77
        },
        "rank": 100,
        "delaytime": 20,             // The time elapsed on the cloud from the end of the audio feed to result acquisition, in ms.
        "wavetime": 1990,          // Duration of the audio file
        "details": [
            {
                "stress": [
                    {
                        "ref": 0,
                        "score": 1,
                        "char": "eh_g"
                    }
                ],
                "char": "egg",
                "dur": 520,         // Word pronunciation duration, in ms.
                "score": 98,       // Pronunciation score
                "phone": [        // Phoneme combination of the word
                    {
                        "char": "e",      // The returned phonemes are not international phonemes. If your product needs to use them, convert them to international phonemes.
                        "score": 96,     // Phoneme score
                        "end": 1460,
                        "start": 1220
                    },
                    {
                        "char": "g",
                        "score": 100,
                        "end": 1740,
                        "start": 1460
                    }
                ],
                "end": 1740,
                "start": 1220
            }
        ],
        "pretime": 23,   // The time taken by the cloud to call the Start interface, in ms.
        "overall": 98
    },
    "eof": 1
}

Response description

Key	Type	Description	Suggestion
recordId	string	The unique ID of the audio file.	For reference
audioUrl	string	The download URL for the user's audio. Audio files are stored for about 20 days by default. For long-term storage, download the files to your own server. The domain name varies by access region and is subject to the actual returned value.	For reference
tokenId	string	The unique ID of the user request.	For reference
applicationId	string	Authorized account and your evaluation AppKey	Internal parameter
dtLastResponse	string	The timestamp of the cloud response.	Internal parameter
eof	int	0: The response is not complete, and more results will follow. 1: All results for this evaluation have been returned.	Internal parameter
refText	string	The input text for evaluation.	For reference
params	object	Echoes the evaluation request parameters passed in by the client.
- app	object
- - applicationId	string	The authorized account, which is the user's AppKey.
- - userId	string	The user ID.
- - timestamp	string	The start time.
- - sig	string	The verification code.
- request	object	The node for the scoring request parameter settings.
- audio	object	The node for the audio parameter settings in the scoring request.
result	object	The evaluation result node.
- version	string	The version of the DPI engine, including the release date.	Internal parameter
- rank	int	The scoring scale.	Internal parameter
- precision	float	The scoring precision.	Internal parameter
- res	string	The name of the resource used for evaluating this task type.	Internal parameter
- forceout	int		Internal parameter
- pretime	int	The time taken by the cloud to call the Start interface, in ms.	Internal parameter
- systime	int	The total time taken for the entire evaluation process, in ms.	Internal parameter
- delaytime	int	The time elapsed on the cloud from the end of the audio feed to result acquisition, in ms.	Internal parameter
- wavetime	int	The audio duration, in ms.	For reference
- overall	float	The overall score.	Recommended field
- pron	float	The pronunciation score.	Internal parameter
- details	array
- - start	int	The start time of the word in the audio, in ms.	For reference
- - end	int	The end time of the word in the audio, in ms.	For reference
- - dur	int	The pronunciation duration of the word, in ms.	For reference
- - char	string	The normalized word text.	For reference
- - score	int	The word score.	For reference
- - fake_pron	int	This field is returned only if the word is not in the dictionary (out-of-vocabulary). To evaluate out-of-vocabulary words, pass the phonemes using the refText parameter.
- - dp_type	int	Indicates normal reading (the dp_type field is not returned), omission (1), or repetition (2). Normal: This field is not returned. Omission: dp_type is 1. Repetition: dp_type is 2.
- - accent	string	This field is not returned if the accent parameter is configured in the request. Otherwise, the field is returned, for example, "accent":"ea".
- - stress	array	Stress pronunciation.
- - - char	string	Phoneme.	For reference
- - - ref	int	Indicates whether the current syllable requires stress. 1: Stress is required. 0: Stress is not required.
- - - score	int	Stress feedback (0 or 1).
- - phone	array	Phoneme-level scoring results. For more information, see Phoneme mapping table.
- - - char	string	The phoneme. If the accent parameter is set in the request, this field contains standard British or American English phonemes. Otherwise, it contains internal mapped phonemes.	For reference
- - - score	int	The phoneme score (0-100).	For reference
- - - phid	int	The position of the phoneme in the word's phonetic transcription.	This field is returned only if phdet is set.
- - - pherr	int	Detects if there is an error in the phoneme (0 or 1).	This field is returned only if phdet is set.
- - - ph2alpha	string	The letter corresponding to the phoneme.	This field is returned only if phdet is set.
- realtime_details	array	Real-time feedback information.
-- char	string	Word.	For reference
-- dp_type	int	0: Normal reading. 1: Omission or not read. 2: Repetition.	For reference
- info	object
- - snr	float	The signal-to-noise ratio (SNR). A higher value indicates clearer audio. Range: 0 to 40 dB. If this parameter affects the score, a corresponding tipId value is set.	For reference
- - clip	float	The audio volume is too high, causing clipping. Range: 0 to 1. If this parameter affects the score, a corresponding tipId value is set.	Internal parameter
- - volume	int	The recording volume. Range: 0 to 180 dB.	For reference
- - tipId	int	Audio quality.	Recommended field
		For more information, see tipId message description.

Suggestions for displaying evaluation results

Dimension	Description	Details
overall	Overall score	The overall evaluation of the word.
pron	Pronunciation score
details.syllable.score	Syllable score	The evaluation of a specific syllable's pronunciation in the word. Check if the field is empty before retrieving the value.
details.stress.score	Word stress score	The evaluation of a specific stressed pronunciation in the word. Check if the field is empty before retrieving the value.
details.phone.score	Phoneme score	The evaluation of a specific phoneme's pronunciation in the word. Check if the field is empty before retrieving the value.
details.dp_type	Word insertion or omission	Indicates whether the word was repeated or omitted. See the response description for the corresponding task type.
details.fake_pron	Voice Spoofing	Indicates whether the word is in the dictionary. If not, a pronunciation is predicted for the word.

Response for a failed evaluation

If the result structure in the response is empty, the evaluation failed.
You can retrieve the `errId` and `error` from the response. Your product can then use the corresponding error code for further processing.

Error codes

For more information about error codes, see Evaluation error codes.