↓
Rivya AI Docs

Rivya Audio Studio: Speech, Sounds and Lyrics

Create speech, dialogue, custom music, sound clips or lyric text, and transcribe a recording. Check each model’s inputs and credit estimate.

Last reviewed on November 11, 2026

Use Audio Studio for speech, scripted dialogue, Suno Sounds or lyric text. Suno Music supports new custom-mode generation. Music extension, WAV conversion and vocal separation from Music results remain unavailable. You can still review existing Music tasks and their status in History, and open or download results while their links remain available. Suno Lyrics returns text and does not start a Music task.

For Suno Music, enter a title and music style. For vocals, enter the words to sing in the lyrics field; for an instrumental, leave lyrics empty. Choose the model version and target duration offered by the form. This does not guarantee exact timing, lyrics or arrangement. Review the model page and credit estimate, then submit once.

You can also start from a public model page after signing in. Studio is not a separate subscription or a prerequisite for every generation; see public pages, sign-in and drafts. A documented model may be paused, so check current availability before preparing a task.

Suno Speech: scenes and supplied scripts

Suno Speech — Suno Speech creates spoken audio with tone, rhythm and emotion. Simple mode writes a script from your scene; Advanced mode reads the script you provide.

Choose Simple to describe an announcement, introduction or spoken scene and let the model write it. Choose Advanced when the wording matters: your main text becomes the script, with optional delivery settings. Switching to Simple drops the Advanced settings.

Suno Speech generates expressive spoken audio and can add background music. For a script with one of the named Zephyr, Puck or Kore voices and WAV output, choose an available Gemini TTS model instead.

Gemini 3.8 Flash TTS · Gemini 3.8 Flash Lite TTS · Gemini 3.1 Flash TTS · Gemini 2.5 Pro TTS

Short-script Gemini speech

The four Gemini TTS models are available for a single speaker reading one short script: 3.8 Flash-Lite, 3.8 Flash, 3.1 Flash and 2.5 Pro. Choose Zephyr, Puck or Kore. The original script is limited to 1,024 UTF-8 bytes, including spaces and line breaks, rather than 1,024 characters. A successful task returns WAV audio.

This path does not accept multiple speakers, attachments, voice cloning or additional scene controls. The reserves are 15, 22, 48 and 48 credits respectively, not promised final charges. Confirmed lower consumption returns the difference; unresolved consumption stays under review. If submission is uncertain, check the existing task instead of submitting it again.

Choose a model by the result you need

Your taskModelWhat to prepare
One speaker reading a scriptElevenLabs Turbo 2.5 or ElevenLabs Multilingual V2The words to be spoken and a voice selection; check language support on the model page
A scripted conversationElevenLabs Dialogue V3Separate dialogue lines, each with text and a voice
A song or instrumental draftSuno MusicA title and music style; lyrics for vocals, or empty lyrics for an instrumental
An ambience, loop or short sound sketchSuno SoundsA sound description and any relevant loop, tempo or key settings
Lyrics before music productionSuno LyricsA lyric brief; the result is text, not a sung track
A recording to transcribeNemotron 3.5 TranscribeUpload one MP3 or WAV; the result is plain text, without word timestamps, translation or video input

ElevenLabs Sound Effect V2 and Audio Isolation are paused; a different plan does not make them available. Suno Sounds generates new sound from a description, not by cleaning an uploaded recording. It has its own controls, so check its model page for the sound-effect task you have in mind. Nemotron 3.5 Transcribe accepts an MP3 or WAV recording and returns text. Rivya does not offer general voice cloning. For work with an existing recording, see audio uploads.

Put the actual spoken words in speech inputs

For text-to-speech, the main text is sent as the material to read aloud. Write the script itself, not a request such as “create a calm 20-second voice-over.” Otherwise the model may speak that instruction. Use the available voice and delivery controls separately.

For example, a short product narration script could be:

Start with your product photo. Choose a background, describe the lighting, and check the result before you download it.

This is an illustrative script, not a tested output or a claim about its duration. Listen to the generated file and adjust the wording or available speed control if it must fit a particular slot. Typing “20 seconds” does not set an exact output length.

Read numbers, names and abbreviations aloud before submitting. For another language, prepare a script that preserves the meaning and sounds natural in that language; changing the site's language does not translate the script. Check the chosen model's language support rather than assuming it matches every Rivya interface language.

For dialogue, add a separate line for each turn and select that line's voice. Fill both fields on every line. Keep speaker directions separate from the words unless the model explicitly supports a notation you intend to use. Chat Studio can help draft or revise text; generating audio remains a separate task.

Describe music and sounds rather than spoken instructions

Use the style field to describe the music, and the lyrics field for the words to sing. Leave lyrics empty for an instrumental. A music brief describes what you want; it does not guarantee exact lyrics, duration or arrangement. Check the model's current availability before planning a new Music task.

For Suno Sounds, describe the sound you need. Use the available loop, tempo and key controls when they matter. Listen across the join to check a loop; the loop setting does not guarantee a seamless result.

For Suno Lyrics, describe the subject and lyric direction. Read and save the returned text before using it in another task. Lyrics do not automatically trigger music generation, and a text-only result is not an audio failure. See music workflows for the distinction between these branches.

Submit once and follow the task

  1. Sign in, open Audio Studio and select an available model. Save a copy of your script and important settings before changing pages or models; unsent work may be lost when you leave.

  2. Complete the selected model's script, dialogue, sound description or lyric brief. Check the required fields and limits, then review the estimated credits. Speech and dialogue costs depend on text length; Sounds and Lyrics have their own task prices. An estimate or reserved amount may differ from the final settled charge. See credits and billing and plans and credit packs.

  3. Submit once. After a task is created, follow its status. If the connection drops or the response is unclear, check the original task in History before creating another. A second request may incur its own charge.

  4. For a task still running, follow queue and waiting guidance. For a recorded failure, consult failed tasks and credit returns. A sound you dislike or a preview that will not play is not by itself a failed generation.

Listen in the context where you will use it

Listen to the whole speech or dialogue result. Check missing words, pronunciation, pauses, voice changes and the intended meaning. A fluent reading can still pronounce a product name incorrectly or omit a key detail.

For video narration or localized speech, compare the audio with the actual scene and timing. Audio Studio does not automatically translate, replace and synchronize a video's spoken track. Prepare the localized wording, generate it, then align and review it in your editing workflow. For more on video narration, see video voice-over and Video Studio.

For existing music results and new sounds, listen beyond the opening: check the ending, unwanted vocals, abrupt changes and loop boundaries where relevant. Confirm that you may use the input and result for your intended purpose; a paid plan is not permission to use someone else's voice or protected material. Read safe upload guidance before sending any source files.

Continue eligible Suno results or save your files

Suno Music supports new custom-mode generation. Music extension, WAV conversion and vocal separation from Music results remain unavailable. Vocal separation is available only for eligible successful Suno Sounds tasks with the required track details. It runs as a separately charged task and cannot be used to upload an arbitrary recording.

If an eligible result offers a follow-up action, check its source and credit notice before confirming. Each accepted follow-up is a separate task with its own credit cost. Follow-up outputs may not offer the same controls, so use the options actually shown. Suno vocal separation does not enable the paused Audio Isolation model.

Save useful audio promptly using output downloads, and save lyric text separately. Keep original scripts, settings and source files in your own project storage; History records submitted tasks. Result links may expire even when the task is still listed. If a preview fails, check the actual file and browser and file troubleshooting before paying for another generation.

For a broader audio project involving voice, music or source recordings, see audio workflows. For podcasts or other spoken content, see the podcast and voice workflow.