Polly is truncating UTF-8 characters when getting speech marks.

[joe1chen]

I'm using an input with the ” character (3-byte UTF-8 E2 80 9D) followed by a period. When calling polly for speech marks json, the response that is returned has truncated the ” character and is only returning the first byte of the unicode character (E2) and dropping the last 2 bytes (80 9D).

To reproduce:

var Polly = new AWS.Polly({
    signatureVersion: 'v4'
});

var params = {
    'Text': "pools”.",
    'OutputFormat': 'json',
    'VoiceId': 'Joanna',
    'SpeechMarkTypes': ['word','sentence']
}

Polly.synthesizeSpeech(params, function (err, data) {
  console.log(data.AudioStream.toString('utf-8'));
}

Output:

{"time":0,"type":"sentence","start":0,"end":6,"value":"pools�"}
{"time":6,"type":"word","start":0,"end":5,"value":"pools"}

The end value, which refers to the end position in the UTF-8 byte stream, also show an end value of 6 in the sentence line, which is incorrect and also indicative that truncation is occurring.

The expected correct output should be:

{"time":0,"type":"sentence","start":0,"end":8,"value":"pools”"}
{"time":6,"type":"word","start":0,"end":5,"value":"pools"}

[AllanZhengYP]

Hi @joe1chen This issue seems like a Service/API issue, which means it does not arise from the SDK. I will forward it to Polly service team to see how we will solve this issue. For now, you can try passing in the readable words only.

Update: This is an issue with Polly service. The team is working on fixing it.

[srchase]

@joe1chen

The Polly Service Team made a fix for this quite a while ago. Closing this issue.

[lock[bot]]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs and link to relevant comments in this thread.