How to Use Universal-2 in opencode
Universal-2 turns recorded interviews, calls, and lectures into transcripts across 99 languages. Compared with Universal-1, it improves recognition of names and alphanumeric strings, plus text formatting.
Prompts to try with Universal-2 in opencode
Copy an example into opencode, or ask in your own words.
Use Universal-2 to transcribe the recorded product demos in this project's media directory. Build a searchable transcript index with links to the matching audio timestamps.
Use Universal-2 to transcribe the English usability interviews in this project, retaining filler words. Create a review page that plays the audio around each selected quote.
Use Universal-2 to transcribe the training recordings in this repository. Generate SRT caption drafts and put them beside the corresponding videos for review.
Use Universal-2 to transcribe this project's two-channel call recordings with multichannel transcription. Preserve channel identifiers in the transcript viewer.
Use Universal-2 to transcribe the recorded architecture discussion in this project. Then draft documentation from the transcript, linking each proposed decision to its audio timestamp.
How to set up Universal-2 in opencode
Pick the way that suits you. Each takes about two minutes.
Use Spiral's MCP server
- Add the server
Leave out --global to add Spiral to the current project only. No header and no client ID or secret: opencode registers itself with Spiral.
opencode mcp add spiral --global --url https://api.spiral.ai/mcp
- Sign in
Your browser opens. Sign in to Spiral, pick the organization and click Authorize, then return to the terminal. No API key. Inside opencode, /mcps starts the same sign-in.
opencode mcp auth spiral
- Check it
Spiral shows as connected.
opencode mcp list
Use the Spiral skill
- Send opencode this message
Your agent reads Spiral's skill and sets itself up.
set up https://spiral.ai/SKILL.md
- Sign in or paste a key
Approve Spiral in the browser, or give the agent a key from the console.
Use the Spiral CLI
- Install and sign in
Installs the Spiral CLI, signs you in and adds its skill to your coding agents.
npx -y spiral-cli@latest init
Call the Spiral API
- Create an API key
In the console, under API keys.
- Run Universal-2
One request; the result comes back by polling or a webhook.
curl https://api.spiral.ai/v1/runs \ -H "Authorization: Bearer $SPIRAL_API_KEY" \ -H "Content-Type: application/json" \ -H "Idempotency-Key: $(uuidgen)" \ -d '{ "provider": "assemblyai", "model": "universal-2", "input": { "audio_url": "<audio_url>" } }'
Then paste one of the prompts above: opencode finds Universal-2 on Spiral, shows the price and runs it.
Universal-2 providers and prices
Spiral runs Universal-2 through AssemblyAI. Open a card for today's price, with every fee included.
$0.195 per hour of audio, per channel. Features, per hour: speaker labels $0.026, key terms $0.065, translation $0.078, speaker identification $0.026, custom formatting $0.039, action items $0.026, summaries $0.026 ($0.091 at medium effort), entity detection $0.104, sentiment $0.026, key phrases $0.013, topic detection $0.195, content moderation $0.195, profanity filter $0.013, PII redaction $0.104, audio redaction $0.065, medical mode $0.195.
Settings you can ask for
Universal-2 works from your audio URL. Ask opencode for any of these in your prompt.
| Setting | Allowed values | Default | Description |
|---|---|---|---|
| Domain | medical-v1 | Enable domain-specific transcription models to improve accuracy for specialized terminology. | |
| Punctuate | on or off | on | Enable Automatic Punctuation, can be true or false |
| Redact pii | on or off | off | Redact PII from the transcribed text using the Redact PII model, can be true or false. |
| Format text | on or off | on | Enable Text Formatting, can be true or false |
| Temperature | 0 to 1 | Control the amount of randomness injected into the model's response. Only takes effect on Universal-3.5 Pro, so Universal-2 ignores it. | |
| Audio end at | a whole number | The point in time, in milliseconds, to stop transcribing in your media file. | |
| Disfluencies | on or off | off | Transcribe Filler Words, like "umm", in your media file; can be true or false. |
| Multichannel | on or off | off | Enable Multichannel transcription, can be true or false. |
| Custom topics | on or off | This parameter does not currently have any functionality attached to it. | |
| Content safety | on or off | off | Enable Content Moderation, can be true or false |
| Iab categories | on or off | off | Enable Topic Detection, can be true or false |
| Redact pii sub | entity_name or hash | The replacement logic for detected PII, can be | |
| Speaker labels | on or off | off | Enable Speaker diarization, can be true or false. Requires |
| Auto highlights | on or off | off | Enable Key Phrases, either true or false |
| Audio start from | a whole number | The point in time, in milliseconds, to begin transcribing in your media file. | |
| Entity detection | on or off | off | Enable Entity Detection, can be true or false |
| Filter profanity | on or off | off | Filter profanity from the transcribed text, can be true or false. |
| Redact pii audio | on or off | off | Generate a copy of the original media file with spoken PII "beeped" out, can be true or false. |
| Speech threshold | 0 to 1 | Reject audio files that contain less than this fraction of speech. | |
| Speakers expected | a whole number | Tells the speaker label model how many speakers it should attempt to identify. | |
| Language detection | on or off | Automatic language detection identifies the spoken language and routes the request to the best model. | |
| Sentiment analysis | on or off | off | Enable Sentiment Analysis, can be true or false. Requires |
| Redact pii audio quality | mp3 or wav | Controls the filetype of the audio created by redact_pii_audio. | |
| Content safety confidence | 25 to 100 | The confidence threshold for the Content Moderation model. | |
| Redact pii return unredacted | on or off | When set to | |
| Language confidence threshold | 0 to 1 | The confidence threshold for the automatically detected language. |
Every API field of Universal-2 Speech to Text is on its tool page: open its card above.
What can you do with Universal-2 in opencode?
Turn project recordings into searchable transcripts, caption drafts, and evidence for documentation. Use Universal-2's word timestamps to link text back to the audio for review.
Interview quote finding
Transcribe recorded interviews, then use word timestamps to locate passages worth quoting. Check selected quotes against the recording before publication.
Podcast caption drafts
Turn spoken podcast episodes into transcripts and timed caption drafts. Review names and technical terms before publishing.
Recorded meeting reference
Create searchable meeting transcripts with speaker labels. Keep a separate, reviewed mapping from speaker labels to people's names.
Lecture study material
Transcribe recorded lectures so an agent can build study notes linked to moments in the recording. Keep the transcript available beside the notes.
Multilingual recording archives
Build a searchable transcript collection from recordings across Universal-2's 99 supported languages, retaining the original recording alongside each transcript.
Two-channel call review
Transcribe calls with each participant recorded on a separate channel. Preserve channel identifiers so reviewers can follow each side of the conversation.
English interview hesitation analysis
Retain filler words in English research interviews when pauses and hesitations matter. Use the transcript to guide a closer listen to relevant passages.
More on Universal-2 and opencode
Overview
Universal-2 in other agents
Questions about Universal-2 in opencode
Can opencode use Universal-2?
Yes. Add Spiral to opencode as an MCP server, and opencode can find and run Universal-2 like any built-in tool.
How much does Universal-2 cost in opencode?
$0.195 per hour of audio, per channel. Features, per hour: speaker labels $0.026, key terms $0.065, translation $0.078, speaker identification $0.026, custom formatting $0.039, action items $0.026, summaries $0.026 ($0.091 at medium effort), entity detection $0.104, sentiment $0.026, key phrases $0.013, topic detection $0.195, content moderation $0.195, profanity filter $0.013, PII redaction $0.104, audio redaction $0.065, medical mode $0.195. You pay from your Spiral balance as you go, with no subscription, and a failed run is not charged. Your opencode plan is separate.
Do I need an AssemblyAI account to use Universal-2 in opencode?
No. Spiral runs Universal-2 on its own AssemblyAI account. You sign in to Spiral once and never paste a key.
How long does it take to set up Universal-2 in opencode?
About two minutes: add the server, then sign in, then check it.
Does opencode wait while Universal-2 runs?
Universal-2 runs in the background. opencode starts the run, checks on it and gives you the json when it's ready.
What is Universal-2 good at?
Use it to turn recorded conversations and spoken presentations into readable text. AssemblyAI's evaluations found improvements over Universal-1 in names, alphanumeric strings, and formatting.
How many languages does Universal-2 support?
Universal-2 supports 99 languages. That makes it useful for transcription collections that span many languages.
How long can a recording be?
The documented transcription limits are 10 hours and 5 GB per file. Split recordings that exceed either limit before transcription.
Can Universal-2 keep hesitations such as um and uh?
For English recordings, you can ask it to retain filler words when hesitations matter to your analysis. Universal-2 removes filler words by default.
When should I preserve separate audio channels?
Preserve them when different speakers were recorded on different channels, such as a two-channel call. Multichannel transcription keeps channel information attached to words and utterances.
Start using Universal-2 in opencode
One Spiral account for every tool your agent needs. No subscription; you pay per run.
Paste this into opencode:
set up https://spiral.ai/SKILL.md


