Skip to main content 首页 创作者 jeremylongshore tons-of-skills-marketplace deepgram-performance-tuning
deepgram-performance-tuning Optimize Deepgram API performance for faster transcription and lower latency.
Use when improving transcription speed, reducing latency,
or optimizing audio processing pipelines.
Trigger: "deepgram performance", "speed up deepgram", "optimize transcription",
"deepgram latency", "deepgram faster", "deepgram throughput".
跳到安装 Skills Marketplace 发现并探索由社区构建的 Agent Skills
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/jeremylongshore/tons-of-skills-marketplace --skill deepgram-performance-tuning命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
下载 Zip 下载中...
langchain-deploy-integration Deploy a LangChain 1.0 / LangGraph 1.0 app to Cloud Run, Vercel, or LangServe correctly — with timeouts sized for chain length, cold-start mitigation, SSE anti-buffering headers, and Secret Manager over .env. Use when prepping a first production deploy, debugging a stream that hangs behind a proxy, or diagnosing p99 latency spikes. Trigger with "langchain deploy", "langchain cloud run", "langchain vercel python", "langchain langserve", or "langchain docker".
jeremylongshore
jeremylongshore/tons-of-skills-marketplace
打开 GitHub 仓库 name deepgram-performance-tuning description Optimize Deepgram API performance for faster transcription and lower latency.
Use when improving transcription speed, reducing latency,
or optimizing audio processing pipelines.
Trigger: "deepgram performance", "speed up deepgram", "optimize transcription",
"deepgram latency", "deepgram faster", "deepgram throughput".
allowed-tools Read, Write, Edit, Bash(ffmpeg:*), Bash(ffprobe:*) version 1.13.0 license MIT author Jeremy Longshore <jeremy@intentsolutions.io> tags ["saas","deepgram","api","performance","optimization"] compatibility Designed for Claude Code
Deepgram Performance Tuning
Overview
Optimize Deepgram transcription performance through audio preprocessing with ffmpeg, model selection for speed vs accuracy, streaming for large files, parallel processing, result caching, and connection reuse. Targets: <2s latency for short files, 100+ files/minute batch throughput.
Performance Levers
Factor Impact Default Optimized Audio format High Any format 16kHz mono WAV Model High nova-3 base (speed) or nova-3 (accuracy) File size High Full file sync Stream >60s, callback >5min Concurrency Medium Sequential 50 parallel (p-limit) Caching Medium None Redis hash by audio+options Features Medium All enabled Disable unused (diarize, utterances)
Instructions
Step 1: Audio Preprocessing with ffmpeg
ffmpeg -i input.mp3 \
-ar 16000 \
-ac 1 \
-acodec pcm_s16le \
-f wav \
output.wav
ffmpeg -i input.wav \
-af "silenceremove=stop_periods=-1:stop_duration=0.5:stop_threshold=-30dB" \
-ar 16000 -ac 1 -acodec pcm_s16le \
trimmed.wav
ffmpeg -i input.wav \
-af "highpass=f=200,lowpass=f=3000,loudnorm=I=-16:TP=-1.5:LRA=11" \
-ar 16000 -ac 1 -acodec pcm_s16le \
clean.wav
import { execSync } from 'child_process' ;
import { statSync } from 'fs' ;
function preprocessAudio (inputPath : string , outputPath : string
originalSize
number
optimizedSize
number
savings
string
const
statSync
size
execSync
`ffmpeg -y -i "${inputPath} " \
-af "silenceremove=stop_periods=-1:stop_duration=0.5:stop_threshold=-30dB,\
highpass=f=200,lowpass=f=3000" \
-ar 16000 -ac 1 -acodec pcm_s16le \
"${outputPath} " 2>/dev/null`
const
statSync
size
const
1
100
toFixed
1
console
log
`Preprocessed: ${inputPath} `
console
log
` Original: ${(originalSize / 1024 ).toFixed(0 )} KB`
console
log
` Optimized: ${(optimizedSize / 1024 ).toFixed(0 )} KB (${savings} % smaller)`
return
Step 2: Model Selection Strategy import { createClient } from '@deepgram/sdk' ;
type Priority = 'accuracy' | 'speed' | 'cost' ;
function selectModel (priority : Priority , audioDuration : number string {
switch (priority) {
case 'accuracy' :
return 'nova-3' ;
case 'speed' :
return audioDuration > 300 ? 'base' : 'nova-2' ;
case 'cost' :
return 'nova-2' ;
default :
return 'nova-3' ;
}
}
function optimizedOptions (priority : Priority return {
model : selectModel (priority, 0 ),
smart_format : true ,
punctuate : true ,
diarize : priority === 'accuracy' ,
utterances : priority === 'accuracy' ,
paragraphs : priority === 'accuracy' ,
summarize : false ,
detect_topics : false ,
sentiment : false ,
};
}
Step 3: Streaming for Large Files import { createClient, LiveTranscriptionEvents } from '@deepgram/sdk' ;
import { createReadStream } from 'fs' ;
async function streamLargeFile (filePath : string Promise <string > {
const deepgram = createClient (process.env .DEEPGRAM_API_KEY !);
const transcripts : string [] = [];
return new Promise ((resolve, reject ) => {
const connection = deepgram.listen .live ({
model : 'nova-3' ,
smart_format : true ,
encoding : 'linear16' ,
sample_rate : 16000 ,
channels : 1 ,
});
connection.on (LiveTranscriptionEvents .Open , () => {
const stream = createReadStream (filePath, { highWaterMark : 32 * 1024 });
stream.on ('data' , (chunk : Buffer {
connection.send (chunk);
});
stream.on ('end' , () => {
connection.finish ();
});
stream.on ('error' , reject);
});
connection.on (LiveTranscriptionEvents .Transcript , (data ) => {
if (data.is_final ) {
const text = data.channel .alternatives [0 ]?.transcript ;
if (text) transcripts.push (text);
}
});
connection.on (LiveTranscriptionEvents .Close , () => {
resolve (transcripts.join (' ' ));
});
connection.on (LiveTranscriptionEvents .Error , reject);
});
}
Step 4: Parallel Batch Processing import pLimit from 'p-limit' ;
import { createClient } from '@deepgram/sdk' ;
async function batchTranscribe (
files : string [],
concurrency = 50 ,
model = 'nova-3'
) {
const client = createClient (process.env .DEEPGRAM_API_KEY !);
const limit = pLimit (concurrency);
const startTime = Date .now ();
const results = await Promise .allSettled (
files.map ((file, i ) =>
limit (async () => {
const fileStart = Date .now ();
const { result, error } = await client.listen .prerecorded .transcribeFile (
require ('fs' ).readFileSync (file),
{ model, smart_format : true , mimetype : 'audio/wav' }
);
if (error) throw error;
const elapsed = Date .now () - fileStart;
console .log (`[${i + 1 } /${files.length} ] ${file} — ${elapsed} ms (${result.metadata.duration} s audio)` );
return { file, result, elapsed };
})
)
);
const totalTime = Date .now () - startTime;
const succeeded = results.filter (r =>status === 'fulfilled' ).length ;
console .log (`\nBatch: ${succeeded} /${files.length} in ${totalTime} ms` );
console .log (`Throughput: ${(files.length / (totalTime / 60000 )).toFixed(1 )} files/min` );
return results;
}
Step 5: Result Caching import { createHash } from 'crypto' ;
import Redis from 'ioredis' ;
const redis = new Redis (process.env .REDIS_URL ?? 'redis://localhost:6379' );
function cacheKey (audioUrl : string , options : Record <string , any >string {
const hash = createHash ('sha256' )
.update (audioUrl + JSON .stringify (options))
.digest ('hex' );
return `dg:cache:${hash} ` ;
}
async function cachedTranscribe (
client : ReturnType <typeof createClient>,
url : string ,
options : Record <string , any >,
ttlSeconds = 3600
) {
const key = cacheKey (url, options);
const cached = await redis.get (key);
if (cached) {
console .log ('Cache hit:' , url.substring (0 , 60 ));
return JSON .parse (cached);
}
const { result, error } = await client.listen .prerecorded .transcribeUrl (
{ url }, options
);
if (error) throw error;
await redis.setex (key, ttlSeconds, JSON .stringify (result));
console .log ('Cached result:' , url.substring (0 , 60 ));
return result;
}
Step 6: Performance Benchmarking async function benchmark (audioUrl : string const client = createClient (process.env .DEEPGRAM_API_KEY !);
const models = ['nova-3' , 'nova-2' , 'base' ] as const ;
console .log ('Performance Benchmark' );
console .log ('=' .repeat (60 ));
for (const model of models) {
const times : number [] = [];
for (let i = 0 ; i < 3 ; i++) {
const start = Date .now ();
const { result, error } = await client.listen .prerecorded .transcribeUrl (
{ url : audioUrl }, { model, smart_format : true }
);
times.push (Date .now () - start);
if (error) { console .error (`${model} error:` , error.message ); break ; }
}
const avg = times.reduce ((a, b ) => a + b, 0 ) / times.length ;
console .log (`${model} : avg ${avg.toFixed(0 )} ms (${times.map(t => `${t} ms` ).join(', ' )} )` );
}
}
Output
Audio preprocessing pipeline (16kHz mono, silence removal, noise reduction)
Model selection strategy by priority (accuracy/speed/cost)
Streaming transcription for large files (>60s)
Parallel batch processing with configurable concurrency
Redis-backed result caching with TTL
Performance benchmarking script
Error Handling Issue Cause Solution Slow transcription Unoptimized audio format Preprocess to 16kHz mono WAV 429 in batch Concurrency too high Reduce p-limit to 50% of plan limit ffmpeg not found Not installed apt install ffmpeg / brew install ffmpegCache stale Audio changed at same URL Include hash of audio content in cache key
Resources 