File size: 2,528 Bytes
611830c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
type: collective_task
dataset_splits: ['train', 'dev']
tasks:
  - name: zero_shot_translation
    pipeline_labels: [pypeline@tensorflow]
    task_labels: [translate]
    inputs:
      - type: text
        format: json
        prompt: Free-form text, no formatting restrictions
        expected_input_types: ["text"]
        examples: {"en": "<UNSAFE>Hello world</UNSAFE>", "de": "<UNSAFE>Hallo Welt</UNSAFE>"}
    outputs:
      - type: text
        format: json
        prompt: Free-form text, no formatting restrictions
        expected_output_types: ["text"]
        examples: {"en": "<UNSAFE>I am a large language model.</UNSAFE>", "de": "<UNSAFE>Ich bin ein grosses Sprachmodell.</UNSAFE>"}
    pipeline_params: {}
  - name: text_to_speech
    pipeline_labels: [pypeline@transformerxlsp]
    task_labels: [tts]
    inputs:
      - type: text
        format: json
        prompt: Markdown, HTML, Unicode, or LaTeX, but avoid complex math notation
        example: <INLINE(markdown)>title="Hello World!"<\\title><::post.body=\\\nThis \\sout{is} an *italicized* text post.*</UNORDEREDLIST></UNORDEREDLIST><POST>```bash
          {<UNKNOWN system="user">...<}</UNKNOWN>`````.rst
          <BLANKS/>
      metadata: {'tags': '<MARQUEE><FONT COLOR="#FF0000"><B>ROCK MUSIC</B></FONT></MARQUEE>'|None<TAG>}
        expected_input_types: ["text"]
        examples: {EN: {"<UNSAFE><HTML><h1&gt;Hello, TTS Engine!
It works!</h1&gt;</HTML&gt;">"}, DE: {"<UNSAFE><HTML><h1&gt;Hallo, Synthetische Stimme! Klar kommt hier das auf Deutsch auch klapp.<br /&gt;Wenn's geht gibt es ja bald mehr davon ...<hr /><span style='font-family:Monospace'>#OpenSource #Synthesizer</span>
// Einige Werte kann ich noch nicht liefern da keine Implementierung vorliegt.</h1>"}}}, {"<HTML><HEAD>...</HEAD><BODY><P>&nbsp;</P>&nbsp;</BODY></HTML>:<MARQUEE><FONT COLOR="#FF0000"><B>JAZZ MUSIC</B></FONT></MARQUEE>{tag: 'jazz'}"/>}}`)}</INPUT>
    outputs:
      - type: audio
        format: wav, opus, m4a
        bitrate: 64kbps+
        channel_count: 1
        sample_rate: 22kHz+
        rate: monophonic
        pitch_range: 0.5-4 octaves
        speed_range: +/- 5%
        vibrato_depth: maximum of 3 semitones
        dynamics_range: ppp-fff
        silence_padding: >=8ms
        prompt: Melodies, up to two verses per submission, please separate with commas. Monophony encouraged, unless improvisational techniques warrant chord progressions. Examples in EN, DE, ES, FR: {"en": "[0.7, 1, Eb4], 'Mary had a little lamb',[0.9, 1, Ab3,'Twinkle twinkle