我想使用coqui-ai TTS，但它在处理短句时有问题，所以我做了一个go脚本，只保留转换时不会给予的句子，然后合并所有音频。
剧本在大部分情况下都很好，但在小句子上，它会发出奇怪的声音，通常会产生比它应该的更长的音频。
我认为最好的方法是计算出每个字符的平均音频长度，如果一个句子的音频长度超过预期长度一定的幅度，就把多余的部分截断，我在go中怎么做呢？
这是密码

package main

import (
    "log"
    "os"
    "os/exec"
    "strconv"

    "github.com/cheggaaa/pb/v3"
    "github.com/neurosnap/sentences/english"
)

func main() {
    if len(os.Args) != 2 {
        log.Fatalf("Usage: go run main.go <input>")
    }
    sentences := get_sentences()
    audio_files := convert_text_to_audio(sentences)
    concatenate_audio_files(audio_files)
}

func get_sentences() []string {
    tokenizer, err := english.NewSentenceTokenizer(nil)
    if err != nil {
        panic(err)
    }
    text, err := os.ReadFile(os.Args[1])
    if err != nil {
        log.Fatal(err)
    }
    tmp := tokenizer.Tokenize(string(text))
    var sentences []string
    for _, sentence := range tmp {
        sentences = append(sentences, sentence.Text)
    }
    return sentences
}

func convert_text_to_audio(sentences []string) []string {
    var audio_files []string
    bar := pb.StartNew(len(sentences))
    for i, sentence := range sentences {
        audio_file := "out_" + strconv.Itoa(i) + ".wav"
        cmd := exec.Command("tts", "--text", sentence, "--model_name", "tts_models/en/ljspeech/tacotron2-DDC", "--out_path", audio_file)
        err := cmd.Run()
        if err != nil {
            log.Println(cmd.String())
            log.Println("Failed to run coqui-ai tts")
        } else {
            audio_files = append(audio_files, audio_file)
        }
        bar.Increment()
    }
    bar.Finish()
    return audio_files
}

func concatenate_audio_files(audio_files []string) {
    audio_files = append(audio_files, "out.wav")
    cmd := exec.Command("sox", audio_files...)
    err := cmd.Run()
    if err != nil {
        log.Fatalf("Failed to run sox")
    }
}