Speech-to-Text

Transcribe audio files with optional speaker diarization.

v0.13.0 API Change

As of v0.13.0, STT methods are accessed via client.STT() instead of client.SpeechToText().

Basic Usage

// Transcribe from URL
result, err := client.STT().TranscribeURL(ctx, "https://example.com/audio.mp3")
if err != nil {
    log.Fatal(err)
}

fmt.Printf("Text: %s\n", result.Text)
fmt.Printf("Language: %s\n", result.LanguageCode)

Transcribe with File Upload

import "github.com/plexusone/elevenlabs-go/stt"

audioData, _ := os.ReadFile("audio.mp3")

result, err := client.STT().Transcribe(ctx, &stt.Request{
    FileBytes: audioData,
    FileName:  "audio.mp3",
    ModelID:   "scribe_v1",
})

Or use the convenience method:

audioData, _ := os.ReadFile("audio.mp3")
result, err := client.STT().TranscribeFile(ctx, audioData)

Note: TranscribeFile takes raw bytes. To transcribe from a file path, read the file first with os.ReadFile().

Speaker Diarization

Identify different speakers in the audio:

result, err := client.STT().TranscribeWithDiarization(ctx, audioURL)
if err != nil {
    log.Fatal(err)
}

for _, word := range result.Words {
    fmt.Printf("[%s] %s (%.2fs - %.2fs)\n",
        word.Speaker, word.Text, word.Start, word.End)
}

Full Options

audioData, _ := os.ReadFile("interview.mp3")

result, err := client.STT().Transcribe(ctx, &stt.Request{
    FileBytes:      audioData,
    FileName:       "interview.mp3",
    ModelID:        "scribe_v1",
    LanguageCode:   "en",           // ISO 639-1 code
    Diarize:        true,           // Enable speaker detection
    TagAudioEvents: true,           // Tag laughter, music, etc.
    NumSpeakers:    2,              // Expected number of speakers
})

Request Options

Option	Type	Description
FileBytes	[]byte	Raw audio file bytes to transcribe
FileName	string	Filename hint (defaults to "audio.wav")
FileURL	string	URL to audio (alternative to FileBytes)
ModelID	string	Transcription model (default: scribe_v1)
LanguageCode	string	ISO 639-1 language code
Diarize	bool	Enable speaker diarization
TagAudioEvents	bool	Tag non-speech audio events
NumSpeakers	int	Expected number of speakers

Response Structure

type Response struct {
    Text         string  // Full transcription text
    LanguageCode string  // Detected language
    Words        []Word  // Word-level timestamps
}

type Word struct {
    Text    string  // The word
    Start   float64 // Start time in seconds
    End     float64 // End time in seconds
    Speaker string  // Speaker ID (if diarization enabled)
}

Use Cases

Meeting Transcription

meetingData, _ := os.ReadFile("meeting.mp3")

result, err := client.STT().Transcribe(ctx, &stt.Request{
    FileBytes:   meetingData,
    FileName:    "meeting.mp3",
    Diarize:     true,
    NumSpeakers: 4,
})

// Group by speaker
speakers := make(map[string][]string)
for _, word := range result.Words {
    speakers[word.Speaker] = append(speakers[word.Speaker], word.Text)
}

Subtitle Generation

result, err := client.STT().TranscribeURL(ctx, videoAudioURL)

// Generate SRT format
for i, word := range result.Words {
    fmt.Printf("%d\n", i+1)
    fmt.Printf("%s --> %s\n", formatTime(word.Start), formatTime(word.End))
    fmt.Printf("%s\n\n", word.Text)
}

Podcast Processing

// Transcribe podcast episode
podcastData, _ := os.ReadFile("episode.mp3")

result, err := client.STT().Transcribe(ctx, &stt.Request{
    FileBytes:      podcastData,
    FileName:       "episode.mp3",
    Diarize:        true,
    TagAudioEvents: true,  // Detect music, laughter, etc.
})

Supported Audio Formats

• MP3
• WAV
• M4A
• FLAC
• OGG
• WEBM

Best Practices

1. Use diarization for multi-speaker content - Interviews, meetings, podcasts
2. Specify language when known for better accuracy
3. Set expected speaker count for more accurate diarization
4. Enable audio event tagging for richer metadata